The data have been split into a test and training set. The training set contains all the calls (over 135,000) that were received from 1 January 2016 to 12 July 2016. You are asked to estimate the number of incoming calls per hour per day from 13 July 2016 to 6 September 2016.

The files you have for download here are:

• train.csv: Every call that was received from 1 January 2016 to 12 July 2016. You will use this file to train your model.
• sample_submission.csv: Your submission CSV file should look like this. Note that the column"time_index" represents ONE HOUR time increments and the column "calls" represents the number of calls received during that one hour time period. The format of time_index is YYYYMMDDHH (year month day hour). For example, the calls associated with time_index 2016071300 would be the calls received between 12:00:00 am until 12:59:59 am on 13 July 2016.
• KenyaPublicHolidays2016.csv: Public holidays in 2016
• NairobiSchoolDates2016.csv: Academic term calendar for 2016
• WeatherNairobi2016.xls: Weather data for Nairobi 2016. The source and explanation of variables can be found here.

Variables in Train.csv:

Each call contains the following fields

Variable Definition

calldate Date (month-day-year) and time of the call

cc_status Case status

maincat Main category call falls into

subcat1 Subcategory call falls under

casepriority Priority of the case

referal Place case referred to

caller_gender Gender of the caller

caller_age Age of the caller

caller_county Area where the call came from

child_age Age of the child in case

child_gender Gender of the child in case

child_county Area where the child is from

parent_age Age of the parent

parent_gender Gender of the parent

parent_county Area where the child is from

Abuser_Relationship Relationship abuser has with the child in case

Neglector_Relationship Relationship neglector has with the child in case

Physical_abuser_Relationship Relationship physical abuser has with the child in case