This exercise simulates a typical situation you’ll face in your job at Apollo. It’s part of your interview and you’ll be evaluated on:
- approach towards solving the business problem
- exploratory data analysis
- feature engineering/selection - candidate should use advanced level of FE & FS techniques:
- PCA
- Anova/Chi
- model development – candidate should use advanced/complex ML algorithms and concepts and compare the performance:
- XG Boost (Extreme Gradient Boosting)
- ADA Boost (Adaptive Boosting)
- ANN using keras/tensorflow/pytorch
- K-Fold Cross Validation
- correct performance metrics
- overall findings
- presentation & communication to technical & non-technical stakeholders
The code you write to build your model should be in PYTHON.
Your thought process and work need to be summarised in a Power Point presentation lasting no more than 15 mins. This is intended to be a technical presentation. Think of it as an end-of-project review with your line manager and other technical experts. In addition to the technical slides, include one or two extra ones summarising your findings & recommendations as you would communicate with non-technical stakeholders.
You are given access to a dataset containing information about Rubber Mixers. Your job is:
- To build a model capable of predicting the efficiency of a mixer. Mixer Efficiency = total_weight/mixing_time
- To identify the features which will have maximum impact on the mixingcycletime.
- To predict the mixer efficiency for the given set of data using advanced ML algorithms and compare the results.