Exponential rolling average & over/under smoothing
In the previous exercise, we saw that our revenue is somewhat flat over time. In this exercise we will dive deeper into the data to see if we can determine why this is the case. We will look at the revenue for a single in-app purchase product we are selling to see if this potentially reveals any trends. As this will have less data then looking at our overall revenue it will be much noisier. To account for this we will smooth the data using an exponential rolling average.
A new daily_revenue dataset has been provided for us, containing the revenue for this product.
This exercise is part of the course
Customer Analytics and A/B Testing in Python
Exercise instructions
- Using the
.ewm()method, calculate the exponential rolling average with aspanof 10 and store it in a columnsmall_scale. - Repeat the previous step, now with a span of 100 and store it in a column
medium_scale. - Finally, calculate the exponential rolling average with a span of 500 and store it in a column
large_scale. - Plot the three averages, along with the raw data. Examine how clear the trend of the data is.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Calculate 'small_scale'
daily_revenue['small_scale'] = daily_revenue.revenue.____(span=____).mean()
# Calculate 'medium_scale'
daily_revenue['___'] = daily_revenue.revenue.____(____=____).____
# Calculate 'large_scale'
daily_revenue[____] = daily_revenue.revenue.____(____=____).____
# Plot 'date' on the x-axis and, our three averages and 'revenue'
# on the y-axis
daily_revenue.plot(x = ____, y =['revenue', 'small_scale', ____, ____])
plt.____