Hi guys,
I am trying to predict the number of conversations my bot will have the next month, everybody are talking about churn but i dont think it does the job. is there something to specify the date like from today to the 28th June i will have n conversations?
I have data from January to make this prediction and i have searched but i can not find anything yet …
i would really appreciate your help.
Hi Safa,
the implementation will very much depend on the way you formulate the problem. In the most natural case you can view it as a time series prediction. For example, “My bot had N_1 conversations last month, N_2 conversations 2 month ago, … N_k conversations k months ago. What is the expected number of conversations in the next month?” In this case you would need more data, as having only 5 data point corresponding to 5 months from January to May will not give you space to build a good model. You might have some additional data from other bots or data at smaller granularity.
As a starting point, you can have a look at several examples of time series analysis available in KNIME:
https://hub.knime.com/knime/workflows/Examples/04_Analytics/07_Time_Series/02_Example_for_Predicting_Time_Series*9SsXq6DBqEsW8q-5
or
https://hub.knime.com/knime/workflows/Examples/50_Applications/10_Energy_Usage/01_Energy_Usage_Time_Series_Prediction*kyzNLnmNn0WWLeaY
Cheers,
Misha
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.