In [1]:
import os
from urllib.request import urlretrieve
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
plt.style.use('seaborn')
In [2]:
#https://stackoverflow.com/questions/11936967/text-file-parsing-with-python
def clean_data(filename):
inputfile = open(filename + '.txt')
outputfile = open(filename + '.csv', 'w')
outputfile.writelines('Date,Temp\n')
for line in inputfile.readlines()[1:]:
outputfile.writelines(','.join(line.split()).replace('99999.9', '') + '\n')
inputfile.close()
outputfile.close()
In [3]:
def get_data(url, filename, force=False):
if force or not os.path.exists(filename + '.txt'):
urlretrieve(url, filename + '.txt')
if force or not os.path.exists(filename + '.csv'):
clean_data(filename)
In [4]:
#http://www.bom.gov.au/climate/change/acorn-sat/#tabs=Data-and-networks
minURL = 'http://www.bom.gov.au/climate/change/acorn/sat/data/acorn.sat.minT.094029.daily.txt'
minFile = 'hobart-min'
get_data(minURL, minFile)
maxURL = 'http://www.bom.gov.au/climate/change/acorn/sat/data/acorn.sat.maxT.094029.daily.txt'
maxFile = 'hobart-max'
get_data(maxURL, maxFile)
In [5]:
minData = pd.read_csv('hobart-min.csv', index_col='Date', parse_dates=True)
maxData = pd.read_csv('hobart-max.csv', index_col='Date', parse_dates=True)
#data = pd.DataFrame.join(maxData, minData, how='outer', lsuffix='_Max', rsuffix="_Min" )
data = minData.merge(maxData, suffixes=('_Min', '_Max'), left_index=True, right_index=True)
data['Temp_Diff'] = data['Temp_Max'] - data['Temp_Min']
In [6]:
data.shape
Out[6]:
In [7]:
data.head()
Out[7]:
In [8]:
data.tail()
Out[8]:
In [9]:
data.describe()
Out[9]:
In [10]:
def apply_common(title=''):
ax.set_ylim(-5,45)
ax.set_title(title)
ax.set_xlabel('Date')
ax.set_ylabel('°Centrigrade')
ax.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
In [11]:
data.hist()
Out[11]:
In [12]:
ax = data.plot()
apply_common('All data')
In [28]:
plt.scatter(data['Temp_Max'], data.index, marker='.')
plt.show()
In [27]:
plt.scatter(data['Temp_Min'], data.index, marker='.')
plt.show()
In [26]:
plt.scatter(data['Temp_Diff'], data.index, marker='.')
plt.show()
In [16]:
filtered_data = data.dropna()
boxplot_data = [filtered_data['Temp_Max'], filtered_data['Temp_Min'], filtered_data['Temp_Diff']]
plt.boxplot(boxplot_data)
plt.xticks([1, 2, 3], ['maximum', 'minimum', 'difference'])
plt.show()
In [17]:
ax = data.resample('m').mean().plot()
apply_common('Monthly mean')
In [18]:
ax = data.resample('d').mean().rolling(365).mean().plot()
apply_common('Rolling 365-day mean')
In [19]:
ax = data.loc['2007-3-1':, 'Temp_Max'].resample('m').mean().plot()
apply_common('Ten year monthly mean')
In [20]:
ax = data.loc['2016-3-1':, 'Temp_Max'].resample('m').mean().plot()
apply_common('One year monthly mean')
In [21]:
ax = data.loc['2007-2-28':, 'Temp_Max'].resample('w').mean().plot()
apply_common('One year weekly mean')
In [22]:
ax = data.loc['2016-3-1':,'Temp_Max'].plot()
apply_common('One year daily')
In [23]:
#https://www.kaggle.com/miguelferia/visualization-of-temperature-data
#http://benalexkeen.com/resampling-time-series-data-with-pandas/
ax = data.resample('AS').mean().plot()
apply_common('Annual summary mean')
In [34]:
data['day#'] = data.index.dayofyear
fig, (ax1, ax2, ax3) = plt.subplots(nrows = 3, ncols = 1)
for key, grp in data.groupby(pd.TimeGrouper(freq='AS'), group_keys=False):
ax1.plot(grp['day#'], grp['Temp_Max'], alpha=0.35)
ax2.plot(grp['day#'], grp['Temp_Min'], alpha=0.35)
ax3.plot(grp['day#'], grp['Temp_Diff'], alpha=0.35)
plt.show()