Lab Assignment 5 - Achu0015/BigDataAna-App-SP2K17 GitHub Wiki

                                            # Lab Assignment 5

Solution 1:

Spark program for Machine Learning task to class images based on categories. Here we used Random forest tree method classification algorithm.

I took the hand written data set images of Digits 0,1 and 4 and letter A. Collected the train and test set data. Using Random Forest Tree classifier we calculate the accuracy and confusion matrix.

The purpose behind taking this data set is to evaluate how well this algorithm predicts the human written digits and letters. But the out come is disappointing, as the accuracy is not very high.

Below are the screen shots of them

The Digit 0 is predicted correctly as zero.

Due to noise in the data set images, some times it does not predict correctly.

Solution 2: Spark client web application for image prediction

Created a client application using Spark API, to predict the images from data set.

Below are the screen shots:

Digit 1 is predicted correctly

Digit 0 is predicted

Whereas Digit 4 is predicted as zero

Solution 3: Google Conversation app through API.AI

I have created a google conversation app for movie suggestion using API.AI and deployed on Google projects.

For the Intent created, several entities are created as shown below

Entity of Actors consist of as below,

We enable actions API on google project and on Integrations tab and then we register our app with unique name and project ID

Once we deploy our application, it will be listed as under review.

I have tried for following basic questions: