Digital demand forecasting

Whether a streaming service, online game, virtual hardware provider or messaging service, being able to correctly predict future demand for the service is vital to being able to ensure there is adequate bandwidth and processing power to support the demand and ensure a quality experience.

Overview

Predict future demand for the service to ensure adequate resources are available.

TARGET

Demand (sessions, meetings, streams, etc.) in the next time period.

Challenge

To start with, the number of connections and users of the session needs to be sourced from the usage database. This information needs to be merged with customer demographics from the CRM along with information about the customers history from the usage database. To more accurately forecast this demand, the data should be combined with financial and web analytics databases along with calendar information. All this data needs to be cleaned, joined and transformed into valuable ML features before going into model training. This pre-modeling prep process can be frustrating and time consuming. We are here to help.

Modeling techniques and libraries

Time series modeling

Simply model the individual demands as a time series.

Package: 
  • Statsmodel.tsa
  • prophet

Machine learning modeling

Model the individual demands using traditional machine learning including time series derived features for demand along with the other features that may influence the future demand.

Package: 
  • Sklearn
  • Tsfresh
  • sktime

Data features

Private or Public Company
CRM
Data Type
Binary
Target
No
Yes
Time since purchase
CRM
Data Type
Continuous
Target
No
Yes
Unique Users per Day
Product Analytics DB
Data Type
Continuous
Target
No
Yes

Related accelerators

No items found.

Get your data science on.