Today’s dashboard project is based on the Boston 311 Service request data from their website here. The data came in a very clean format so no work was required in Alteryx. I decided to challenge myself by creating a machine learning model to predict whether a service request would be resolved on time or not. Firstly, I tried to integrate R into Tableau but was unable to due to some unknown issue. This is a problem that may be able to be resolved with more time but I was able to find a workaround. I coded the classification model on R studio. See a snapshot of the model below.
Since I was not able to integrate R into tableau directly, I had to hard code the values. I focused on 3 main metrics: the police district, the reason for the report, and the department which dealt with the report. I was able to calculate the probability of where a service would be completed on time or not. The average target duration and average duration of resolutions were not good metrics due to some significant outliers that the model was able to pick up on. Here is a snapshot of the dashboard. See my tableau public to see more.
