This notebook shows how to plot multiindexed data (like the pipeline output for multiple assets) in three different ways.
In [95]:
import numpy as np
import matplotlib.pyplot as plt
from quantopian.pipeline import Pipeline, CustomFactor
from quantopian.pipeline.data import Fundamentals
from quantopian.research import run_pipeline
from quantopian.pipeline.filters import StaticAssets
from quantopian.research import symbols
from quantopian.pipeline.data import USEquityPricing
import matplotlib.pyplot as plt
import pandas as pd
In [104]:
assets_screen = StaticAssets(symbols(['NKE', 'UAA']))
fundamentals = {
'total_assets': Fundamentals.total_assets.latest,
'total_liabilities': Fundamentals.total_liabilities.latest,
'market_cap': Fundamentals.market_cap.latest
}
my_pipeline = Pipeline(columns=fundamentals, screen=assets_screen)
data = run_pipeline(my_pipeline, '2002-1-01', '2018-6-27', 90)
In [105]:
data.head()
Out[105]:
In [106]:
# Slice dataframe to get values for each stock
idx = pd.IndexSlice
data_1 = data.loc[idx[:, symbols('NKE')], :]
data_2 = data.loc[idx[:, symbols('UAA')], :]
2. Similar idea, cleaner code¶
In [107]:
# This constructs the same plot, but using cleaner code
columns = list(data.columns)
dates = data.index.levels[0]
# Iterate through columns and plot each one
for column in columns:
plt.plot(dates, data_1[column], label='NKE '+str(column))
plt.plot(dates, data_2[column], label='UAA '+str(column))
plt.legend()
plt.title('NKE & UAA Balance Sheet Comparison (in 100s of billions)')
plt.show()
3. Using QGrid¶
In [9]:
import qgrid
# Get data in Qgrid
qgrid_widget = qgrid.show_grid(data)
qgrid_widget
In [10]:
# Filter the second index column to show only rows for T,
# then save the changed dataframe
T_data = qgrid_widget.get_changed_df()
In [11]:
# Filter the second index column to show only rows for VZ,
# then save the changed dataframe
VZ_data = qgrid_widget.get_changed_df()
In [12]:
T_data.head()
Out[12]:
In [13]:
VZ_data.head()
Out[13]:
In [14]:
# Now plot using the same code from above
columns = list(data.columns)
dates = data.index.levels[0]
plt.clf() # clears previous plot
# Iterate through columns and plot each one
for column in columns:
plt.plot(dates, T_data[column], label='T '+str(column))
plt.plot(dates, VZ_data[column], label='VZ '+str(column))
plt.legend()
plt.title('Historical T CashFlow (in billions)')
plt.show()
