weather predictions

data

We have collected an historical series of weather data: weather_h

data collected from darkskay api

Data contains a series of features, some are categorical, some continous.

objective

We want to predict the event of rain.

Please note the difference between prediction and forecast and explain how you would run a forecast instead of a prediction.

tasks

To guide the execution we write a check list on few useful steps

• data quality
• outlier - missing
• feature distribution
• feature transformation
• dimension reduction
• target variable definition
• feature importance
• model selection
• prediction - scoring

execution

It’s expected around 4h for the data analysis part plus the documentation/presentation

evaluation

The quality of the final report will be evaluated together with the arguments to the discussion and data visualization. In case a good prediction accuracy won’t be reached it’s important to describe what are the expectations and what else can be done to improve the results.

In case of too high performances please review the feature selection.

delivery

Written report and used sources (in any preferred language).

CRM

How would you design a metric that is both business and data driven, closely aligned with revenue targets and based on CRM data?

Describe your approach and how you would build the metric considering the information you could access from a standard CRM database.

SQL

Connect to dauvi.org/phpsql/ with

• user: guest
• pass: aTgera45(]

and write a query calculating the sum of the compensation times the number of calls for each agent from the tables demo_agent and demo_history