SWAPNIL TRIPATHI

 SUMMER INTERNSHIP-Study of Various Units of Refinery 

Organization: Indian Oil corporation ltd - IOCL 

Duration:  May`18 - Jul`18 

Description: The training involved the visits to various units of refining namely OHCU,DHDT,DHDS,HGU1 & HGU2. The project emphasizes on crude refining process in various units, its products and residue. 

 1.Title: HR Analytics 

Problem Statement: We have 3 months of employee attendance data of ATLIQ hardware. The requirement of HR was about : 

• 1. Working preferences of employees between Work From Home and Work From Office & its reason
• 2. Percentage of overall sick leave to monitor employee wellness.

Outcomes: 

• Developed a Dashboard for Presence, Work From Home(WFH), and Sick Leave(SL) etc of the employees.
• As June approaches, presence % decreases ; WFH increases ;SL are increasing may be due to hot summers.
• Most people are present on Monday or Tuesday. And Most people take WFH on Friday or Thursday.

2. Title: Psychographic Segmentation 

Problem Statement:To find out psychographics of the respondents and then understand each segment by the usage of the products to get understanding of the cluster physiography enable them to custom products according to the needs of users. 

Description: 

• Data collected from 1082 students of B-schools and other professional courses.
• The consumers were asked about ownership and behavior of products. 14 psychographic statements, each asked to be rated on a 11-point-scale.

 Outcomes: 

• 3 Clusters were formed using K-means Clustering in Python.
• Clusters were named as Purposeful,Inactive & Spendthrift. Clustering and segment linkages drawn demographics and product ownership.
• Brand recommendations for each cluster.

3.Title: Instagram User Analytics. 

Problem Statement: 

• To derive useful insights from the Instagram user's data /metadata about User behaviour & about inactive users & marketing opportunities.
• Top trends/hashtag & regarding Bots & fake accounts etc

 4.Title: Credit Risk Prediction using Machine Learning

 Problem Statement: To predict if a person will default the credit or not based on various predictors. 

Description: 

• Various attributes like age,loan amount,interest rate,loan as percentage if income,previous credit history were used for building ML models.
• Models Used are KNN Classifier, Decision Tree & Random Forest Classifier

Outcome: 

• Revenue is not independent of Budget and metascore
• The two predictors capture/explain 60% of variability of revenue.

5. Title: Calorie burnt prediction using Machine Learning 

Problem Statement:  

• To find Correlation between various predictors & target variable.
• To predict number of calories you would burn during a specific duration using certain features.

Description: 

• Predictor variables taken here are UserID,Gender,Age,Height,Weight,Duration,Heart Rate,Body Temperature. Target variable is 'Calories' burnt.
• Models used: XGBoost Regressor

Outcome:Duration & Heart rate;Duration & body temperature;Duration & Calorie burnt are highly correlated. Heart rate & Calories burnt are highly correlated. 

6. Title: EDA & Regression Analysis On Happiness Score

Problem Statement: To identify trends in happiness index for various countries & identify factors determining state of happiness & their impacts for period of 2018 to 2022.Also ,we will do a regression analysis of 2022 dataset. 

Outcome: 

• Happier countries tend to be those with longer life expectancies, and a higher GDP.
• GDP per Capita, Healthy Life Expectancy and Social Support are 3 major factor that determining country's state of Happiness

 7. Title: Analyzing Cryptocurrencies using Power BI

 Objective: 

• Building interactive dashboard of various cryptocurrencies using slicer over the years to showcase their historical data.
• Showcasing historical market cap over years & Forecast their market cap for the future.

• Singing
• Trekking
• Table Tennis
• Cricket
• International relations
• Geopolitics
• Travelling