Shojib_DS_Portfolio

A high level overview about my works and projects

Project 1 estimate which employesses are more likely to quit

Supervised learning model, Classification Problem

• Create a tool for HR team to help them identify the employeees who are quitting.
• Used Logistic Regression and Ensemble model(Random Forest),on top to Scikitlearn.
• Deep Learning model,trained on the top of Keras API, to observe which model works well.
• used ARTIFICIAL NEURAL NETWORKS architecture

Project 2 Forecasting the future sales of chain store

Supervised learning, Regression

• Build a Machine Learning model to predict their future sales based on past data.
• build on the top of Facebook Prophet framework.

Project 3 Segmenting the customers into three distinct groups so better marketing largeted add can launch.

Unsupervised Technique

• separating customers three distinct groups.In order to better targeted marketing campaign can launch
• KMeans Algorithm used.
• Choose cluster by using Elbow-method
• PCA also used reducing dimension
• APPLY AUTOENCODERS (PERFORM DIMENSIONALITY REDUCTION USING AUTOENCODERS)

Project 4 Classify chest disease to help doctor to diagnosis

A image classification trained Machine Learning Model on the top of Keras API.

Images classification model

• Model takes four types of image Healthy, Covid-19, Bacterial Pneumonia , Viral Pnemonia.
• Model use Convolutinal Neural Network Architecture.
• Applying Transfer Learning technique with the help of a pretraied model ResNet50.

Project 5 Sensitivity analysis using Natural Language Processing techniques

Natural Language Processing

• Analyze customers reviews towards a Amazon great Product Alexa.
• As it is text data so data pre-processing ,like remove punctuation and stopwords ,vectorization,
• Model use NAIVE BAYES CLASSIFIER to train and predict the data.