Python HTML Reports in Python/v3

How to make HTML reports with Python, Pandas, and Plotly Graphs.

This page in another language

Note: this page is part of the documentation for version 3 of Plotly.py, which is not the most recent version.
See our Version 4 Migration Guide for information about how to upgrade.

Generate HTML reports with D3 graphs
using Python, Plotly, and Pandas

d3.js is an amazing JavaScript library for creating interactive, online graphics and charts. Plotly lets you create d3.js charts using Python, R, or MATLAB. This IPython notebook shows you how to embed these charts in an HTML report that you can then share by email or host on a website.

Once you've created your report generation script, you can automate it with a task scheduler like cron or Windows Task Manager, and email it to yourself and your team.

import plotly as py import pandas as pd import numpy as np from datetime import datetime from datetime import time as dt_tm from datetime import date as dt_date import plotly.plotly as py import plotly.tools as plotly_tools import plotly.graph_objs as go import os import tempfile os.environ['MPLCONFIGDIR'] = tempfile.mkdtemp() from matplotlib.finance import quotes_historical_yahoo import matplotlib.pyplot as plt from scipy.stats import gaussian_kde from IPython.display import HTML 

Step 1: Generate 2 graphs and 2 tables¶

First graph: 2014 Apple stock data with moving average¶

Let's grab Apple stock data using the matplotlib finance model from 2014, then take a moving average with a numpy convolution.

x = [] y = [] ma = [] def moving_average(interval, window_size): window = np.ones(int(window_size))/float(window_size) return np.convolve(interval, window, 'same') date1 = dt_date( 2014, 1, 1 ) date2 = dt_date( 2014, 12, 12 ) quotes = quotes_historical_yahoo('AAPL', date1, date2) if len(quotes) == 0: print "Couldn't connect to yahoo trading database" else: dates = [q[0] for q in quotes] y = [q[1] for q in quotes] for date in dates: x.append(datetime.fromordinal(int(date))\ .strftime('%Y-%m-%d')) # Plotly timestamp format ma = moving_average(y, 10) 

Now graph the data with Plotly. See here for Plotly's line plot syntax and here for getting started with the Plotly Python client.

xy_data = go.Scatter( x=x, y=y, mode='markers', marker=dict(size=4), name='AAPL' ) # vvv clip first and last points of convolution mov_avg = go.Scatter( x=x[5:-4], y=ma[5:-4], \ line=dict(width=2,color='red',opacity=0.5), name='Moving average' ) data = [xy_data, mov_avg] py.iplot(data, filename='apple stock moving average') 

Save the plot URL - we'll use it when generating the report later.

first_plot_url = py.plot(data, filename='apple stock moving average', auto_open=False,) print first_plot_url 

https://plotly.com/~jackp/1841

Second graph: Scatter matrix of 2014 technology and CPG stocks¶

Let's use the Pandas package and Plotly subplots to compare different tech. and CPG stocks in 2014.
This graph was inspired by this IPython notebook and GitHub user twiecki.

tickers = ['AAPL', 'GE', 'IBM', 'KO', 'MSFT', 'PEP'] prices = [] for ticker in tickers: quotes = quotes_historical_yahoo(ticker, date1, date2) prices.append( [q[1] for q in quotes] ) 

We have all the stock prices in a list of lists - use the code snippet below to convert this into a Pandas dataframe.

df = pd.DataFrame( prices ).transpose() df.columns = tickers df.head() 

Use Plotly's get_subplots() routine to generate an empty matrix of 6x6 subplots. We'll fill these in by plotting all stock ticker combinations against each other (ie, General Electric stock versus Apple stock)

fig = plotly_tools.get_subplots(rows=6, columns=6, print_grid=True, horizontal_spacing= 0.05, vertical_spacing= 0.05) 

This is the format of your plot grid! [31] [32] [33] [34] [35] [36] [25] [26] [27] [28] [29] [30] [19] [20] [21] [22] [23] [24] [13] [14] [15] [16] [17] [18] [7] [8] [9] [10] [11] [12] [1] [2] [3] [4] [5] [6]

def kde_scipy(x, x_grid, bandwidth=0.4, **kwargs): """Kernel Density Estimation with Scipy""" # From https://jakevdp.github.io/blog/2013/12/01/kernel-density-estimation/ # Note that scipy weights its bandwidth by the covariance of the # input data. To make the results comparable to the other methods, # we divide the bandwidth by the sample standard deviation here. kde = gaussian_kde(x, bw_method=bandwidth / x.std(ddof=1), **kwargs) return kde.evaluate(x_grid) subplots = range(1,37) sp_index = 0 data = [] for i in range(1,7): x_ticker = df.columns[i-1] for j in range(1,7): y_ticker = df.columns[j-1] if i==j: x = df[x_ticker] x_grid = np.linspace(x.min(), x.max(), 100) sp = [ go.Histogram( x=x, histnorm='probability density' ), \ go.Scatter( x=x_grid, y=kde_scipy( x.as_matrix(), x_grid ), \ line=dict(width=2,color='red',opacity='0.5') ) ] else: sp = [ go.Scatter( x=df[x_ticker], y=df[y_ticker], mode='markers', marker=dict(size=3) ) ] for ea in sp: ea.update( name=' vs '.format(x_ticker,y_ticker),\ xaxis='x<>'.format(subplots[sp_index]),\ yaxis='y<>'.format(subplots[sp_index]) ) sp_index+=1 data += sp # Add x and y labels left_index = 1 bottom_index = 1 for tk in tickers: fig['layout']['xaxis<>'.format(left_index)].update( title=tk ) fig['layout']['yaxis<>'.format(bottom_index)].update( title=tk ) left_index=left_index+1 bottom_index=bottom_index+6 # Remove legend by updating 'layout' key fig['layout'].update(showlegend=False,height=1000,width=1000, title='Major technology and CPG stock prices in 2014') fig['data'] = data py.iplot(fig, height=1000, width=1000, filename='Major technology and CPG stock prices in 2014 - scatter matrix')