Garden and Lawn mowers review amazon predictive

December 14, 2019
Data cleaning was done in exccel and the cleaned data was then imported to AzureML Studio by name of "amazon text nlp exam.csv" Question4. The datastructure should be column1 - User ID (Whom to Recommend) column2 - Item ID (What to Recommend) Column3 - Rating[integar] (On what basis to be rated) For better accuracy in association user features and item features data can be included. Question1. Multiclass neural network with default parameters giving 59.86% Accuracy Question2. Multiclass neural net with random sweep tuning 64.39% Accuracy(100 iterations) Multiclass neural net with entire grid tuning 64.96% Accuracy(100 iterations) Multiclass neural net with entire grid tuning 64.45% Accuracy(25 iterations) Question3. Multiclass neural net is giving some error when implemented with hidden layer(some Microsoft lib error) So I have used two class neural net supported by hyper parameter tuning(entire grid) with 2 hidden layers(auto) - 65% Accuracy hyper parameter tuning(entire grid) with 2 hidden layers(200 neurons) - 64.9% Accuracy