Welcome to the second project of the Machine Learning Engineer Nanodegree! In this notebook, some template code has already been provided for you, and it will be your job to implement the additional functionality necessary to successfully complete this project. Sections that begin with 'Implementation' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Please specify WHICH VERSION OF PYTHON you are using when submitting this notebook. Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode.
In this project, you will employ several supervised algorithms of your choice to accurately model individuals' income using data collected from the 1994 U.S. Census. You will then choose the best candidate algorithm from preliminary results and further optimize this algorithm to best model the data. Your goal with this implementation is to construct a model that accurately predicts whether an individual makes more than $50,000. This sort of task can arise in a non-profit setting, where organizations survive on donations. Understanding an individual's income can help a non-profit better understand how large of a donation to request, or whether or not they should reach out to begin with. While it can be difficult to determine an individual's general income bracket directly from public sources, we can (as we will see) infer this value from other publically available features.
The dataset for this project originates from the UCI Machine Learning Repository. The datset was donated by Ron Kohavi and Barry Becker, after being published in the article "Scaling Up the Accuracy of Naive-Bayes Classifiers: A Decision-Tree Hybrid". You can find the article by Ron Kohavi online. The data we investigate here consists of small changes to the original dataset, such as removing the 'fnlwgt' feature and records with missing or ill-formatted entries.
Run the code cell below to load necessary Python libraries and load the census data. Note that the last column from this dataset, 'income', will be our target label (whether an individual makes more than, or at most, $50,000 annually). All other columns are features about each individual in the census database.
In [12]:
# Import libraries necessary for this project
import numpy as np
import pandas as pd
from time import time
from IPython.display import display # Allows the use of display() for DataFrames
# Import supplementary visualization code visuals.py
import visuals as vs
# Pretty display for notebooks
%matplotlib inline
# Load the Census dataset
data = pd.read_csv("census.csv")
# Success - Display the first record
display(data.head(n=5))
the function is written during one of the Kaggle challenge by one of it's participants. [https://github.com/sayalidzambre/allstate-claims-severity/blob/master/skutils.py] to show the data's properties
In [33]:
import skutils
skutils.pretty_stats(data)
In [65]:
features_con = list(data.select_dtypes(exclude=['object', 'category']).columns)
print features_con
features_num = list(data.select_dtypes(include=['object','category']).columns)
print features_num
In [66]:
import matplotlib.pyplot as plt
import seaborn as sns
f = plt.figure(figsize=(15,15))
for i in range(len(features_con)):
ax = f.add_subplot(len(features_con),3,3*i+1)
sns.distplot(data[features_con[i]])
ax = f.add_subplot(len(features_con),3,3*i+2)
sns.distplot(data[features_con[i]])
ax = f.add_subplot(len(features_con),3,3*i+3)
sns.violinplot(y=data[features_con[i]],x=data.income)
In [67]:
#sns.set(style="ticks", color_codes=True)
#iris = sns.load_dataset("iris")
sns.pairplot(data)#,kind="reg"
Out[67]:
A cursory investigation of the dataset will determine how many individuals fit into either group, and will tell us about the percentage of these individuals making more than \$50,000. In the code cell below, you will need to compute the following:
'n_records''n_greater_50k'.'n_at_most_50k'.'greater_percent'. HINT: You may need to look at the table above to understand how the 'income' entries are formatted.
In [68]:
# TODO: Total number of records
test=data["age"]
n_records = len(test)
# TODO: Number of records where individual's income is more than $50,000
import re
test2=data["income"]
n_greater_50k=[i for i in test2 if re.findall(r">",i)] #data[data.income==">50K"].income.count()
n_greater_50k= len(n_greater_50k)
# TODO: Number of records where individual's income is at most $50,000
n_at_most_50k = [i for i in test2 if re.findall(r"<",i)] #data[data.income=="<=50K"].income.count()
n_at_most_50k = len(n_at_most_50k)
# TODO: Percentage of individuals whose income is more than $50,000
#from __future__ import division
greater_percent = float(n_greater_50k)*100/n_records
# Print the results
print "Total number of records: {}".format(n_records)
print "Individuals making more than $50,000: {}".format(n_greater_50k)
print "Individuals making at most $50,000: {}".format(n_at_most_50k)
print "Percentage of individuals making more than $50,000: {:.2f}%".format(greater_percent)
Featureset Exploration
Before data can be used as input for machine learning algorithms, it often must be cleaned, formatted, and restructured — this is typically known as preprocessing. Fortunately, for this dataset, there are no invalid or missing entries we must deal with, however, there are some qualities about certain features that must be adjusted. This preprocessing can help tremendously with the outcome and predictive power of nearly all learning algorithms.
A dataset may sometimes contain at least one feature whose values tend to lie near a single number, but will also have a non-trivial number of vastly larger or smaller values than that single number. Algorithms can be sensitive to such distributions of values and can underperform if the range is not properly normalized. With the census dataset two features fit this description: 'capital-gain' and 'capital-loss'.
Run the code cell below to plot a histogram of these two features. Note the range of the values present and how they are distributed.
In [69]:
# Split the data into features and target label
income_raw = data['income']
features_raw = data.drop('income', axis = 1)
# Visualize skewed continuous features of original data
vs.distribution(data)
For highly-skewed feature distributions such as 'capital-gain' and 'capital-loss', it is common practice to apply a logarithmic transformation on the data so that the very large and very small values do not negatively affect the performance of a learning algorithm. Using a logarithmic transformation significantly reduces the range of values caused by outliers. Care must be taken when applying this transformation however: The logarithm of 0 is undefined, so we must translate the values by a small amount above 0 to apply the the logarithm successfully.
Run the code cell below to perform a transformation on the data and visualize the results. Again, note the range of values and how they are distributed.
In [70]:
# Log-transform the skewed features
skewed = ['capital-gain', 'capital-loss']
features_log_transformed = pd.DataFrame(data = features_raw)
features_log_transformed[skewed] = features_raw[skewed].apply(lambda x: np.log(x + 1))
# Visualize the new log distributions
vs.distribution(features_log_transformed, transformed = True)
In addition to performing transformations on features that are highly skewed, it is often good practice to perform some type of scaling on numerical features. Applying a scaling to the data does not change the shape of each feature's distribution (such as 'capital-gain' or 'capital-loss' above); however, normalization ensures that each feature is treated equally when applying supervised learners. Note that once scaling is applied, observing the data in its raw form will no longer have the same original meaning, as exampled below.
Run the code cell below to normalize each numerical feature. We will use sklearn.preprocessing.MinMaxScaler for this.
In [71]:
# Import sklearn.preprocessing.StandardScaler
from sklearn.preprocessing import MinMaxScaler
# Initialize a scaler, then apply it to the features
scaler = MinMaxScaler() # default=(0, 1)
numerical = ['age', 'education-num', 'capital-gain', 'capital-loss', 'hours-per-week']
features_log_minmax_transform = pd.DataFrame(data = features_log_transformed)
features_log_minmax_transform[numerical] = scaler.fit_transform(features_log_transformed[numerical])
# Show an example of a record with scaling applied
display(features_log_minmax_transform.head(n = 5))
From the table in Exploring the Data above, we can see there are several features for each record that are non-numeric. Typically, learning algorithms expect input to be numeric, which requires that non-numeric features (called categorical variables) be converted. One popular way to convert categorical variables is by using the one-hot encoding scheme. One-hot encoding creates a "dummy" variable for each possible category of each non-numeric feature. For example, assume someFeature has three possible entries: A, B, or C. We then encode this feature into someFeature_A, someFeature_B and someFeature_C.
| someFeature | someFeature_A | someFeature_B | someFeature_C | ||
|---|---|---|---|---|---|
| 0 | B | 0 | 1 | 0 | |
| 1 | C | ----> one-hot encode ----> | 0 | 0 | 1 |
| 2 | A | 1 | 0 | 0 |
Additionally, as with the non-numeric features, we need to convert the non-numeric target label, 'income' to numerical values for the learning algorithm to work. Since there are only two possible categories for this label ("<=50K" and ">50K"), we can avoid using one-hot encoding and simply encode these two categories as 0 and 1, respectively. In code cell below, you will need to implement the following:
pandas.get_dummies() to perform one-hot encoding on the 'features_raw' data.'income_raw' to numerical entries.0 and records with ">50K" to 1.
In [72]:
# TODO: One-hot encode the 'features_log_minmax_transform' data using pandas.get_dummies()
import pandas as pd
features_final = pd.get_dummies(features_raw) #
#print features_final
# TODO: Encode the 'income_raw' data to numerical values
income = pd.get_dummies(income_raw) #income_raw.apply(lambda x: 1 if x == ">50K" else 0)
income = income_raw.apply(lambda x: 1 if x == ">50K" else 0 )
#income = np.where(income_raw == '<=50K', 0, 1)
#print income
# Print the number of features after one-hot encoding
encoded = list(features_final.columns)
print "{} total features after one-hot encoding.".format(len(encoded))
# Uncomment the following line to see the encoded feature names
#print encoded
Now all categorical variables have been converted into numerical features, and all numerical features have been normalized. As always, we will now split the data (both features and their labels) into training and test sets. 80% of the data will be used for training and 20% for testing.
Run the code cell below to perform this split.
In [73]:
# Import train_test_split
from sklearn.cross_validation import train_test_split
# Split the 'features' and 'income' data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(features_final,
income,
test_size = 0.2,
random_state = 0)
# Show the results of the split
print "Training set has {} samples.".format(X_train.shape[0])
print "Testing set has {} samples.".format(X_test.shape[0])
CharityML, equipped with their research, knows individuals that make more than \$50,000 are most likely to donate to their charity. Because of this, *CharityML* is particularly interested in predicting who makes more than \$50,000 accurately. It would seem that using accuracy as a metric for evaluating a particular model's performace would be appropriate. Additionally, identifying someone that does not make more than \$50,000 as someone who does would be detrimental to *CharityML*, since they are looking to find individuals willing to donate. Therefore, a model's ability to precisely predict those that make more than \$50,000 is more important than the model's ability to recall those individuals. We can use F-beta score as a metric that considers both precision and recall:
$$ F_{\beta} = (1 + \beta^2) \cdot \frac{precision \cdot recall}{\left( \beta^2 \cdot precision \right) + recall} $$In particular, when $\beta = 0.5$, more emphasis is placed on precision. This is called the F$_{0.5}$ score (or F-score for simplicity).
Looking at the distribution of classes (those who make at most \$50,000, and those who make more), it's clear most individuals do not make more than \$50,000. This can greatly affect accuracy, since we could simply say "this person does not make more than \$50,000" and generally be right, without ever looking at the data! Making such a statement would be called naive, since we have not considered any information to substantiate the claim. It is always important to consider the naive prediction for your data, to help establish a benchmark for whether a model is performing well. That been said, using that prediction would be pointless: If we predicted all people made less than \$50,000, CharityML would identify no one as donors.
Accuracy measures how often the classifier makes the correct prediction. It’s the ratio of the number of correct predictions to the total number of predictions (the number of test data points).
Precision tells us what proportion of messages we classified as spam, actually were spam. It is a ratio of true positives(words classified as spam, and which are actually spam) to all positives(all words classified as spam, irrespective of whether that was the correct classificatio), in other words it is the ratio of
[True Positives/(True Positives + False Positives)]
Recall(sensitivity) tells us what proportion of messages that actually were spam were classified by us as spam. It is a ratio of true positives(words classified as spam, and which are actually spam) to all the words that were actually spam, in other words it is the ratio of
[True Positives/(True Positives + False Negatives)]
For classification problems that are skewed in their classification distributions like in our case, for example if we had a 100 text messages and only 2 were spam and the rest 98 weren't, accuracy by itself is not a very good metric. We could classify 90 messages as not spam(including the 2 that were spam but we classify them as not spam, hence they would be false negatives) and 10 as spam(all 10 false positives) and still get a reasonably good accuracy score. For such cases, precision and recall come in very handy. These two metrics can be combined to get the F1 score, which is weighted average(harmonic mean) of the precision and recall scores. This score can range from 0 to 1, with 1 being the best possible F1 score(we take the harmonic mean as we are dealing with ratios).
'accuracy' and 'fscore' to be used later.Please note that the the purpose of generating a naive predictor is simply to show what a base model without any intelligence would look like. In the real world, ideally your base model would be either the results of a previous model or could be based on a research paper upon which you are looking to improve. When there is no benchmark model set, getting a result better than random choice is a place you could start from.
HINT:
In [76]:
import numpy as np
from sklearn.metrics import accuracy_score,recall_score,fbeta_score
TP = np.sum(income) # Counting the ones as this is the naive case. Note that 'income' is the 'income_raw' data
#encoded to numerical values done in the data preprocessing step.
print TP
FP = income.count() - TP # Specific to the naive case
#print FP
income_prediction = income.apply(lambda x:1)
TN = 0 # No predicted negatives in the naive case
FN = 0 # No predicted negatives in the naive case
#FN = sum(map(lambda x,y : 1 if x==1 and y==0 else 0, income,income_prediction))
#TP = sum(map(lambda x,y : 1 if x==1 and y==1 else 0, income,income_prediction))
#FP = sum(map(lambda x,y : 1 if x==0 and y==1 else 0, income,income_prediction))
TP=float(TP)
FP=float(FP)
print FN,TP,TN,FP
# TODO: Calculate accuracy, precision and recall
accuracy = float(TP)/float(TP+FP) #income.count()[1] 1.0 * n_greater_50k / (n_greater_50k + n_at_most_50k)
recall = float(TP)/float(TP+FN)#1.0 * n_greater_50k / (n_greater_50k)
precision = float(TP)/float(TP+FP) # tp/tp+fp
#print TP,FP,income.count()[1]
#print "accuracy = {} ,recall={} ,precision={}".format (accuracy,recall,precision)
# TODO: Calculate F-score using the formula above for beta = 0.5 and correct values for precision and recall.
# HINT: The formula above can be written as (1 + beta**2) * (precision * recall) / ((beta**2 * precision) + recall)
beta = 0.5
fscore = (1 + beta**2) * (precision * recall) / ((beta**2 * precision) + recall)
# Print the results
#print type(accuracy)
print "Naive Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]".format(accuracy, fscore)
The following are some of the supervised learning models that are currently available in scikit-learn that you may choose from:
List three of the supervised learning models above that are appropriate for this problem that you will test on the census data. For each model chosen
HINT:
Structure your answer in the same format as above^, with 4 parts for each of the three models you pick. Please include references with your answer.
Answer: 1- Gaussian Naive Bayes: 2- SVM: 3- Ensemble Method, Adaboost: 4- Gradient Boosting
1- Naive Bayes classifier is a supervised linear classifier, and works based on probabilistic behavior of the training sets. Most of the time, for the numeric implementations K-Nearest Neighbors and K-Means clustering algorithms can be implemented. Naive Bayes classifier is best used for classifying emails, texts, symbols, and names. Naive Bayes classification predicts the probability of each class based on the feature vector, for text classification of continious big data with prior distribution of data. Its a good fit for this data becasue there is a clear separation of the classes, and since its easy to implement, its a good idea to apply it on this data to get a base line. It is simple to implement, becasue it assumes that all the features are independent from each other. and this allows for fast classification to train quickly with less training sets. Its weakness is that it assumes all the fatures are independent from each other which not always true. [ref: https://medium.com/@gp_pulipaka/applying-gaussian-na%C3%AFve-bayes-classifier-in-python-part-one-9f82aa8d9ec4][http://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.GaussianNB.html][https://www.analyticsvidhya.com/blog/2017/09/naive-bayes-explained/] [http://blog.aylien.com/naive-bayes-for-dummies-a-simple-explanation/] 2- SVM is widely applied in the biological and other sciences. SVM has been used to classify proteins with up to 90% of the compounds classified correctly[ https://en.wikipedia.org/wiki/Support_vector_machine]. SVM is a supervised machine learning algorithm which can be used for classification or regression problems. It uses a technique called the kernel trick to transform the data and based on these transformations it finds an optimal boundary between the possible outputs. Simply put, it does some extremely complex data transformations, then figures out how to seperate the data based on the labels or outputs of classes.Its best used for multi-dimentional datasets, where there is a clear gap between the classes. But for large data it is time consuming to classify complex domains of data. It can overfit the data if the data is noisy or classes overlap. If there is overlap/noise, Naive Bayes would be better. Since the number of features are not large and there may be a clear margin of seperation SVM can be applied to the current data. Also this doesn’t work well for large data sets as training time will be cubic of the size of the data set. i.e. lot of data and lots of features, SVM is not preferable, [http://www.yaksis.com/posts/why-use-svm.html][https://en.wikipedia.org/wiki/Support_vector_machine][scikit-learn.org/stable/modules/svm.html] [scikit-learn.org/stable/modules/generated/sklearn.svm.SVC.html]
3- The real-world application in industry where Decision Trees with Ensemble Methods (Bagging, AdaBoost, Random Forest, Gradient Boosting) can be applied is that it can be used in oil and gas industry where a company should decide if it is worth drilling on a test sight or not [https://www.slb.com/~/media/Files/resources/oilfield_review/ors00/win00/p2_9.pdf]. Boosting is a general ensemble method that creates a strong classifier from a number of weak classifiers. This is done by building a model from the training data, then creating a second model that attempts to correct the errors from the first model. Models are added until the training set is predicted perfectly or a maximum number of models are added. AdaBoost is best used to boost the performance of decision trees on binary classification problems. It can be used to boost the performance of any machine learning algorithm but it is best used with weak learners. These are models that achieve accuracy just above random chance on a classification problem. The most suited and therefore most common algorithm used with AdaBoost are decision trees with one level. Because these trees are so short and only contain one decision for classification It performs very fast compared to SVM, and less proning to overfitting. On the other hand, noisy data and outliers could impact the performance of this mehod. Its a good option for this project because the data is not noisy and not very large. [http://scikit-learn.org/stable/modules/ensemble.html#adaboost,http://scikit-learn.org/stable/modules/ensemble.html][https://machinelearningmastery.com/boosting-and-adaboost-for-machine-learning/][https://www.analyticsvidhya.com/blog/2015/11/quick-introduction-boosting-algorithms-machine-learning/]
4- Gradient Boosting is one of the most efficient alogrithims. One application of it is in Higgs Boson Discovery from the Large Hadron Collider dataset, here physicits can extract signal of Higgs Boson particle from background noises. [http://proceedings.mlr.press/v42/chen14.pdf] Gradient Boosting Classifier is fast trainng without sacrificing accuracy,can handle different types of predictor variables and accomodate missing data Its weekness ia that its unable to compute conditional class probabilites and suffers from long sequential computation times. Reasons for being a good Candidate: its quite powerful and it is more expressive as it builds one tree at a time and corrects errors made by previously trained trees.
To properly evaluate the performance of each model you've chosen, it's important that you create a training and predicting pipeline that allows you to quickly and effectively train models using various sizes of training data and perform predictions on the testing data. Your implementation here will be used in the following section. In the code block below, you will need to implement the following:
fbeta_score and accuracy_score from sklearn.metrics.X_test, and also on the first 300 training points X_train[:300].beta parameter!
In [77]:
# TODO: Import two metrics from sklearn - fbeta_score and accuracy_score
from sklearn.metrics import fbeta_score, accuracy_score
def train_predict(learner, sample_size, X_train, y_train, X_test, y_test):
'''
inputs:
- learner: the learning algorithm to be trained and predicted on
- sample_size: the size of samples (number) to be drawn from training set
- X_train: features training set
- y_train: income training set
- X_test: features testing set
- y_test: income testing set
'''
results = {}
# TODO: Fit the learner to the training data using slicing with 'sample_size' using .fit(training_features[:], training_labels[:])
start = time() # Get start time
learner.fit(X_train[:sample_size],y_train[:sample_size])
end = time() # Get end time
# TODO: Calculate the training time
results['train_time'] = end - start
# TODO: Get the predictions on the test set(X_test),
# then get predictions on the first 300 training samples(X_train) using .predict()
start = time() # Get start time
predictions_test = learner.predict(X_test)
predictions_train = learner.predict(X_train[:300])
end = time() # Get end time
# TODO: Calculate the total prediction time
results['pred_time'] = end - start
# TODO: Compute accuracy on the first 300 training samples which is y_train[:300]
results['acc_train'] = accuracy_score(y_train[:300],predictions_train.round())
# TODO: Compute accuracy on test set using accuracy_score()
results['acc_test'] = accuracy_score(y_test,predictions_test.round())
# TODO: Compute F-score on the the first 300 training samples using fbeta_score()
results['f_train'] = fbeta_score(y_train[:300],predictions_train.round(),beta=0.5)
# TODO: Compute F-score on the test set which is y_test
results['f_test'] = fbeta_score(y_test,predictions_test.round(),beta=0.5)
# Success
print "{} trained on {} samples.".format(learner.__class__.__name__, sample_size)
# Return the results
return results
In the code cell, you will need to implement the following:
'clf_A', 'clf_B', and 'clf_C'.'random_state' for each model you use, if provided.'samples_1', 'samples_10', and 'samples_100' respectively.Note: Depending on which algorithms you chose, the following implementation may take some time to run!
In [86]:
# TODO: Import the three supervised learning models from sklearn
from sklearn.naive_bayes import GaussianNB
from sklearn.svm import SVC
from sklearn.ensemble import GradientBoostingClassifier
# TODO: Initialize the three models
clf_A = GaussianNB()
clf_B = SVC(random_state = 0)
clf_B = GradientBoostingClassifier(n_estimators=100, learning_rate=0.1, max_depth=3, random_state=50)
clf_C = AdaBoostClassifier(random_state=0)
# TODO: Calculate the number of samples for 1%, 10%, and 100% of the training data
# HINT: samples_100 is the entire training set i.e. len(y_train)
# HINT: samples_10 is 10% of samples_100
# HINT: samples_1 is 1% of samples_100
samples_100 = len(y_train)
samples_10 = len(y_train)/10
samples_1 = len(y_train)/100
# Collect results on the learners
results = {}
for clf in [clf_A, clf_B, clf_C]:
clf_name = clf.__class__.__name__
results[clf_name] = {}
for i, samples in enumerate([samples_1, samples_10, samples_100]):
results[clf_name][i] = \
train_predict(clf, samples, X_train, y_train, X_test, y_test)
# Run metrics visualization for the three supervised learning models chosen
vs.evaluate(results, accuracy, fscore)
In this final section, you will choose from the three supervised learning models the best model to use on the student data. You will then perform a grid search optimization for the model over the entire training set (X_train and y_train) by tuning at least one parameter to improve upon the untuned model's F-score.
HINT:
Look at the graph at the bottom left from the cell above(the visualization created by vs.evaluate(results, accuracy, fscore)) and check the F score for the testing set when 100% of the training set is used. Which model has the highest score? Your answer should include discussion of the:
Answer:
From the above it appears GradientBoosting classifier Ensemble Method seems to be most appropriate for the task of identifying individuals that make more than \$50,000. Although it takes a little alonger to train, it gives a higher accuracy and f-score compared to Ada Boost Classifier and Gaussian NB.
HINT:
When explaining your model, if using external resources please include all citations.
Answer: The GradientBoosting classifier is chosen to be the final model. The bottom center figure shows that it has arond 0.8 accuracy score on the testing set, which means 80% of the time the model predicts correctly that the testing data has income more than 50k. GradientBoosting classifier Ensemble Method is a types of Tree model. A Tree model is, in layman’s terms, like the Twenty Questions guessing game. The guesser might have questions like “Is it bigger than a bread box?”, “Is it alive?”, etc. The size or lifeness of the thing being guessed at is a “feature”. By winnowing down what is likely or unlikely based on these questions, you end up with a likely (but possibly wrong) answer. Part of the strategy in 20 questions is to order the questions correctly: the first few questions should be broad, so as to eliminate large number of possibilities. The last few questions should be more specific to hone in on the “best” possible answer. Now, what happens when a Tree ML is trained on the data set, the algorithm tries to come up with a set of “questions” that are ‘optimal”. Unfortunately, there is no perfect solution. So, there are different strategies to try to build the Tree ML. The idea behind GBM is a more sophisticated. Roughly, the idea is to again, combine weak predictors. The trick is to find areas of misclassification and then “boost” the importance of those incorrectly predicted data points. And repeat. The output for the new tree is then added to the output of the existing sequence of trees in an effort to correct or improve the final output of the model.A fixed number of trees are added or training stops once loss reaches an acceptable level or no longer improves on an external validation dataset. When we receive a new donor, the new donor is parsed through all set of questions and based on the answers will arrive at the final step and final prediction. The way it works is that we trained the model with 80% of the dataset and the model learned based on the given features, if the individual data has more than 50k income or not. Then we tested this model on the 20% of the data which the model hasn't seen before; and we measured the accuracy and f-score. The accuracy is the measurement of corrected predictions over the total predictions. recall or sensitivity is the ratio of correct predictions over correct predictions and wrong predictions. f-score is another form of measurement of accuracy performance where both accuracy and sensitivity of the model are included in. for a perfect model, f-score is 1.0 and worst model is 0. For this example, the f-score is around 0.8. The GradientBoosting classifier, divides the training datasets into smaller sets and learn based on their features and if the model does not predict the remaing subset of the training set correctly, it will put more weight on them as if they are more important compared to the rest of training set, and it will learn those subsets again. It will go through this loop until it converges.
Fine tune the chosen model. Use grid search (GridSearchCV) with at least one important parameter tuned with at least 3 different values. You will need to use the entire training set for this. In the code cell below, you will need to implement the following:
sklearn.grid_search.GridSearchCV and sklearn.metrics.make_scorer.clf.random_state if one is available to the same state you set before.parameters = {'parameter' : [list of values]}.max_features parameter of your learner if that parameter is available!make_scorer to create an fbeta_score scoring object (with $\beta = 0.5$).clf using the 'scorer', and store it in grid_obj.X_train, y_train), and store it in grid_fit.Note: Depending on the algorithm chosen and the parameter list, the following implementation may take some time to run!
In [80]:
# TODO: Import 'GridSearchCV', 'make_scorer', and any other necessary libraries
from sklearn import grid_search, cross_validation
from sklearn.metrics import make_scorer
from sklearn.cross_validation import StratifiedKFold
'''
#StratifiedKFold
'''
# TODO: Initialize the classifier
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=0.1,
max_depth=3,
random_state=50)
# TODO: Create the parameters list you wish to tune
parameters = {'n_estimators' : [100,500,1000],
'learning_rate' : [0.01,0.5,1],
'max_depth' : [3,10,50,100]}
# TODO: Make an fbeta_score scoring object
scorer = make_scorer(fbeta_score, beta=0.5)
#splitss=(y_train.count())/100 #slipt the y_train in chunks of y_train/100
crossv = StratifiedKFold(y_train, n_folds=2 , random_state = 50) #n_splits=3, test_size=0.5, random_state=0
# TODO: Perform grid search on the classifier using 'scorer' as the scoring method
grid_obj = grid_search.GridSearchCV(clf, param_grid=parameters, cv = crossv, scoring=scorer, n_jobs=-1, verbose=10)
#n_jobs=-1 : the computation will be dispatched on all the CPUs of the computer.
# TODO: Fit the grid search object to the training data and find the optimal parameters
grid_fit = grid_obj.fit(X_train, y_train)
# Get the estimator
best_clf = grid_fit.best_estimator_
# Make predictions using the unoptimized and model
predictions = (clf.fit(X_train, y_train)).predict(X_test)
best_predictions = best_clf.predict(X_test)
# Report the before-and-afterscores
print "Unoptimized model\n------"
print "Accuracy score on testing data: {:.4f}".format(accuracy_score(y_test, predictions))
print "F-score on testing data: {:.4f}".format(fbeta_score(y_test, predictions, beta = 0.5))
print "\nOptimized Model\n------"
print "Final accuracy score on the testing data: {:.4f}".format(accuracy_score(y_test, best_predictions))
print "Final F-score on the testing data: {:.4f}".format(fbeta_score(y_test, best_predictions, beta = 0.5))
'''# TODO: Import 'GridSearchCV', 'make_scorer', and any other necessary libraries
from sklearn.metrics import make_scorer
from sklearn.grid_search import GridSearchCV
# TODO: Initialize the classifier
clf = AdaBoostClassifier(random_state=0)
# TODO: Create the parameters list you wish to tune, using a dictionary if needed.
# HINT: parameters = {'parameter_1': [value1, value2], 'parameter_2': [value1, value2]}
parameters = {"n_estimators": [10,40,70,100],"learning_rate":[.1,.5,1,1.5],
"loss":[ "linear", "square", "exponential"]}
# TODO: Make an fbeta_score scoring object using make_scorer()
scorer = make_scorer(fbeta_score,beta=0.5)
# TODO: Perform grid search on the classifier using 'scorer' as the scoring method using GridSearchCV()
grid_obj = GridSearchCV(clf,param_grid=parameters,scoring=scorer)
y_train = y_train#.round()
X_train = X_train#.round()
print type(y_train), y_train[:5]
print type(X_train), X_train[:5]
# TODO: Fit the grid search object to the training data and find the optimal parameters using fit()
grid_fit = grid_obj.fit(X_train,y_train)
# Get the estimator
best_clf = grid_fit.best_estimator_
# Make predictions using the unoptimized and model
predictions = (clf.fit(X_train, y_train)).predict(X_test)
best_predictions = best_clf.predict(X_test)
# Report the before-and-afterscores
print "Unoptimized model\n------"
print "Accuracy score on testing data: {:.4f}".format(accuracy_score(y_test, predictions))
print "F-score on testing data: {:.4f}".format(fbeta_score(y_test, predictions, beta = 0.5))
print "\nOptimized Model\n------"
print "Final accuracy score on the testing data: {:.4f}".format(accuracy_score(y_test, best_predictions))
print "Final F-score on the testing data: {:.4f}".format(fbeta_score(y_test, best_predictions, beta = 0.5))'''
Out[80]:
In [82]:
grid_fit.best_params_
Out[82]:
Note: Fill in the table below with your results, and then provide discussion in the Answer box.
Answer:
Optimized models's accuracy and the f-score are 0.8668 and 0.7440 which are higher than their counter parts in Unoptimized Mode. The reason is the good efficency of cross-validation techniques to find the the model with the better papameters compared to a unoptimized model's parameter. The Naive Bayes benchmark in Q1 ahd accuracy and f-score of 0.2478 and 0.2917, which are very low. So the learning models and cross validations worked perfectly to imporve the outcome.
Stratification works by keeping the balance or ratio between labels/targets of the dataset. So if your whole dataset has two labels (e.g. Positive and Negative) and these have a 30/70 ratio, and you split in 10 subsamples, each stratified subsample should keep the same ratio. Reasoning: Because machine-learned model performance in general are very sensitive to a sample balancing, using this strategy often makes the models more stable for subsamples.
Split vs Random Split. A split is just a split, usually for the purpose of having separate training and testing subsamples. But taking the first X% for a subsample and the remaining for another subsample might not be a good idea because it can introduce very high bias. There is where random split comes into play, by introducing randomness for the subsampling.
K-fold Cross-validation vs Random Split. K folds comprises in creating K subsamples. Because you now have a more significant number of samples (instead of 2), you can separate one of the subsamples for testing and the remaining subsamples for training, do this for every possible combination of testing/training folds and average the results. This is known as cross-validation. Doing K-fold cross-validation is like doing a (not random) split k times, and then averaging. A small sample might not benefit from k-fold cross-validation, while a large sample usually always do benefit from cross-validation. A random split is a more efficient (faster) way of estimating, but might be more prone to sampling bias than k-fold cross-validation. Combining stratification and random split is an attempt to have an effective and efficient sampling strategy that preserves label distribution.
An important task when performing supervised learning on a dataset like the census data we study here is determining which features provide the most predictive power. By focusing on the relationship between only a few crucial features and the target label we simplify our understanding of the phenomenon, which is most always a useful thing to do. In the case of this project, that means we wish to identify a small number of features that most strongly predict whether an individual makes at most or more than \$50,000.
Choose a scikit-learn classifier (e.g., adaboost, random forests) that has a feature_importance_ attribute, which is a function that ranks the importance of features according to the chosen classifier. In the next python cell fit this classifier to training set and use this attribute to determine the top 5 most important features for the census dataset.
When Exploring the Data, it was shown there are thirteen available features for each individual on record in the census data. Of these thirteen records, which five features do you believe to be most important for prediction, and in what order would you rank them and why?
Answer:
I think the top most influential features are:
1- capital-loss
2- capital-gain
3- age
4- occupation
5- education-num
The capital-loss and capital-gain are directly related to the income of the individual. I ranked them as number 1 and 2, because the large values of capital-loss and capital-gain indicate that the person's income is usually high. Then I ranked age as third, because older people usually have higher income. occupation is ranked 4th because the pay-per-hour is very much come from the occupation. and education-num if fifth as it indicated how much leverage the person would have in the job market and hence income.
Choose a scikit-learn supervised learning algorithm that has a feature_importance_ attribute availble for it. This attribute is a function that ranks the importance of each feature when making predictions based on the chosen algorithm.
In the code cell below, you will need to implement the following:
'.feature_importances_'.
In [83]:
# TODO: Import a supervised learning model that has 'feature_importances_'
from sklearn.ensemble import AdaBoostClassifier
# TODO: Train the supervised model on the training set using .fit(X_train, y_train)
model = GradientBoostingClassifier(n_estimators=100,
learning_rate=0.1,
max_depth=3,
random_state=50)
model.fit(X_train, y_train)
# TODO: Extract the feature importances using .feature_importances_
importances = model.feature_importances_
# Plot
vs.feature_plot(importances, X_train, y_train)
Observe the visualization created above which displays the five most relevant features for predicting if an individual makes at most or above \$50,000.
Answer: They are pretty close to what I had in mind, except the Married-civ-spouse. it shows that the number of education years, has the highest weight of 0.6, so it would be the most important feature, and then following that, are age, Married-civ-spouse, capital-loss and capital-gain, respectibely with lower weights.
Edited: I think Marriage status ended up being a strong feature because intuitively as people get married they start to take making money and saving money more seriosly becasue they want to build a great life with their spouse.
How does a model perform if we only use a subset of all the available features in the data? With less features required to train, the expectation is that training and prediction time is much lower — at the cost of performance metrics. From the visualization above, we see that the top five most important features contribute more than half of the importance of all features present in the data. This hints that we can attempt to reduce the feature space and simplify the information required for the model to learn. The code cell below will use the same optimized model you found earlier, and train it on the same training set with only the top five important features.
In [85]:
# Import functionality for cloning a model
from sklearn.base import clone
# Reduce the feature space
X_train_reduced = X_train[X_train.columns.values[(np.argsort(importances)[::-1])[:5]]]
X_test_reduced = X_test[X_test.columns.values[(np.argsort(importances)[::-1])[:5]]]
# Train on the "best" model found from grid search earlier
clf = (clone(best_clf)).fit(X_train_reduced, y_train)
# Make new predictions
reduced_predictions = clf.predict(X_test_reduced)
# Report scores from the final model using both versions of data
print "Final Model trained on full data\n------"
print "Accuracy on testing data: {:.4f}".format(accuracy_score(y_test, best_predictions))
print "F-score on testing data: {:.4f}".format(fbeta_score(y_test, best_predictions, beta = 0.5))
print "\nFinal Model trained on reduced data\n------"
print "Accuracy on testing data: {:.4f}".format(accuracy_score(y_test, reduced_predictions))
print "F-score on testing data: {:.4f}".format(fbeta_score(y_test, reduced_predictions, beta = 0.5))
Answer:
The accuracy and f-score on the reduced data are decreased. It's mainly because we are feeding the learners less features to learn from so it will be less accurate, with having less information on the whole features of the dataset. f-score is decreased by around 0.01 amount, its not noticable because both accuracy and recall affect the f-score. But the accuracy score is decreaded by 0.4, which is more noticable, the reduction is due to reasons mentioned above, having less features, hence less accurate.
Note: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to
File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.