Recommendation systems are not only useful for better source of driving traffic to retail or online businesses. It is also useful to reach the individual and give them a customized experience irrespective of their change in their interests over time.

Thus, not only does it capture the attention and increases the return on investment, it also increases the customer satisfaction which directly leads to loyal customers who generate profit.

Problem Statement:

Analytics Vidhya is a platform which also conducts a lot of online and live hackathons. Thus it becomes a place where a lot of programmers work on various topics and upload their codes.

Since the users are on different levels in their programming skills, the goal of this program is to automatically predict or identify the number of attempts the user is likely to solve a code with reasonable accuracy, given the problems details and the specialisation of the user. The data contains the description about the type of the problem, the difficulty level of the problem and also to identify the user is in beginner or an expert and their field of expertise in ML, dynamic programming or graph algorithms etc.

Doing so, the programming committee will be able to identify where and when the user will get stuck during coding and thus, will be able to the users individually by giving relevant hints or suggestions to the problems they face automatically.

• The Notebook file is Recommendation Engine.ipynb.
• The output file is submission.csv.

To know how to solve the problem statement Approach

We have obtained a good f1_score score for test data. The thresholds for each of the model has helped in decent split and we have successfully achieved the objective

The models can be tuned for hyperparameter optimization, but because the training data is large, it takes time for parametrs to get tuned.

This dataset was part of Recommendation Engine conducted my Analytics Vidhya, for more info check:Link to Competition