Deep Learning Model for Stress Detection
There are a plenty of wearable devices such as Apple watch, Fitbit Watch and other wearable technologies. Those wearable devices monitor our health, activities, health conditions, use that information to provide customized recommendations for us to follow good and healthy habits and at the same time stopping or reducing the unhealthy and bad habits. Stress is one of the silent killer in this century, as long as we can monitor and detect stress and it’s behaviour and patterns, we can find out solutions and medications to handle it. In this case study the objective is to understand the dataset provided and organize the data in such a way to represent the information to anyone. Apply the deep learning models to predict stress and classify the results.
Deploy ML Model for Fraudulent Transactions Identification in credit card spends
In the following dataset there are 492 fraudulent transactions, which is 0.172% of the total transactions. Try to identify those 492 transactions using the outlier detection methods and report out the accuracy, that is how many fraudulent transactions your outlier detection technique can successfully detect out of total 492. The objective is to deploy the trained model, hence generate the model trained in the form of PMML script, Pickle Dump and Joblib dump. Using the above format, you should be able to upload any dataset and generate predictions for any new transaction.
Advertisement Recommendation from internet traffic
This use cases involves recommendation of advertisement on internet pages, whether a page has advertisement present or not. If no advertisement is present, can we potentially identify those pages and recommend to relevant advertisers given the fact that they have many visitors to their website. It is important to generate the recommendations when the customer is active and live on the web page and the recommendation generation process should not take much time and the length of stay on a particular web page is very less. Build the entire recommendation model using different algorithms and verify the best algorithm that fits to the scenario.
Calorie Prediction using Neural Network Model
Epicurious - Recipes with Rating and Nutrition is a dataset available over Kaggle, Recipes from Epicurious by rating, nutritional content, and categories have been provided. The objective here is to analyze the data, work on feature engineering and predict the calories based on the content of the recipe, for example given a list of ingredients your model should be able to predict how much calorie it contains, this is especially of importance to the nutritionists. For the restaurant owners it is also important to understand the user’s choice and preference towards any particular recipe. Hence create another model predicting the rating that customers would like to provide based on the content of the recipe (title).
Note: Only use JSON format data, DO NOT use CSV format
Neural Network for Classification Model
In the case study the insurance company has given Information about customers consists of 86 variables and includes product usage data and socio-demographic data derived from zip area codes. The data was supplied by the Dutch data mining company Sentient Machine Research and is based on a real world business problem. The training set contains over 5000 descriptions of customers, including the information of whether or not they have a caravan insurance policy. A test set contains 4000 customers of whom only the organizers know if they have a caravan insurance policy.