In [ ]:

    

%load_ext autoreload
%autoreload 2
import boystats as B
import pandas as pd
import matplotlib
import matplotlib.pyplot as plt
import seaborn as sns
import collections
import numpy as np
sns.set_context('talk')


    

In [ ]:

    

B.convert('data', 2020)


    

In [ ]:

    

df = pd.read_csv('data/volume.csv')
df['date'] = pd.to_datetime(df.date)
plt.figure(figsize=(14, 8))
df.plot(x='date', y='volume', style='k', alpha=.2, legend=False, ax=plt.gca())
plt.plot(df.date, df.volume.ewm(com=7).mean(), color='k')
plt.ylabel('Volume /ml')
plt.title('Feed volume');


    

In [ ]:

    

df = pd.read_csv('data/sick.csv', dtype=dict(time=str))
df['date'] = pd.to_datetime(df.date)
df['count'] = 1
f = df.groupby('date')['count'].sum()
f = f.reindex(pd.date_range(f.index.min(), f.index.max()), fill_value=0).reset_index().rename(columns=dict(index='date'))

plt.figure(figsize=(14, 8))
plt.plot(f['date'], f['count'].ewm(com=14).mean(), color='k')
f.plot(x='date', y='count', style='k', alpha=.2, legend=False, ax=plt.gca())
plt.ylabel('Daily vomit frequency')
plt.title('Vomits over time')
plt.ylim(f['count'].min(), f['count'].max())

events = pd.read_csv('data/events.csv', converters=dict(date=pd.to_datetime))
plt.vlines([d.toordinal() for d in events.date], f['count'].min(), f['count'].max(), linestyles='dashed')
for _, row in events.iterrows():
    plt.annotate(row.event, (row.date.toordinal(), f['count'].max()), va='top', textcoords='offset points', xytext=(5, -2), fontsize=18)


    

In [ ]:

    

plt.figure(figsize=(14,8))
hours = df.time[~df.time.isna()].apply(lambda x: int(x[:2]))
counts = collections.Counter(hours)
xs = np.arange(0, 24)
plt.gca().bar(xs, [counts.get(x, 0) for x in xs])
plt.gca().set_xticks(xs)
plt.gca().set_xticklabels(['{:02d}'.format(x) for x in xs])
plt.title('Vomits by hour-of-day')

plt.figure(figsize=(14,8))
days = df.date.apply(lambda x: x.day_name())
counts = collections.Counter(days)
xlabels = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']
xs = list(range(len(xlabels)))
plt.gca().bar(xs, [counts.get(x, 0) for x in xlabels])
plt.gca().set_xticks(xs)
plt.gca().set_xticklabels(xlabels)
plt.title('Vomits by day-of-week');