Data Analyst & Statistician

Janak Rudani

Turning data into decisions

Statistics student at the University of Manitoba. I build end-to-end analytical pipelines in R and Python regression, time series, classification and spend my free time building PCs, chasing bugs, and listening to jazz.

View Projects ↓ GitHub ↗
6
Portfolio Projects
R · Py · SQL
Languages
0.84
Best AUC Score
R · tidyverse · ggplot2 · forecast · caret Python · pandas · scikit-learn · statsmodels · pmdarima SQL · Window Functions · Self-Joins · CTEs Streamlit · Plotly · Interactive Dashboards Linear · Logistic · Poisson GLM · SARIMA ANOVA · ANCOVA · AUC · AIC · Hypothesis Testing R · tidyverse · ggplot2 · forecast · caret Python · pandas · scikit-learn · statsmodels · pmdarima SQL · Window Functions · Self-Joins · CTEs Streamlit · Plotly · Interactive Dashboards Linear · Logistic · Poisson GLM · SARIMA ANOVA · ANCOVA · AUC · AIC · Hypothesis Testing

Featured Work

Projects

01 / 06

Telco Customer Churn
Logistic Regression

Binary classification on 7,043 telecom customers to predict churn and quantify its drivers. Odds ratios reveal month-to-month contracts and fiber optic service as the strongest risk factors backed by Type II likelihood-ratio tests.

AUC ≈ 0.84 in both R and Python

RPython Logistic Regression ROC / AUCOdds Ratios

github.com/Jrudani21/telco-churn-logistic-regression ↗

02 / 06

DC Bike Share Demand
Count Regression

Modelled 731 days of Capital Bikeshare data using Poisson and Negative Binomial GLMs. Detected severe overdispersion (variance/mean = 357×) and used IRRs to quantify weather and seasonal effects on daily rentals.

NB model AIC ~1,200 better than Poisson

RPython Poisson / NB GLM IRRUCI Dataset

github.com/Jrudani21/bike-share-demand-count-regression ↗

03 / 06

Air Passengers
Time Series Forecasting

Full Box-Jenkins SARIMA pipeline log transform, stationarity testing (ADF/KPSS), ACF/PACF analysis, 36-month forecast with prediction intervals, and head-to-head comparison against Facebook Prophet.

SARIMA MAPE ~3–5% vs Prophet ~4–6%

RPython SARIMA ProphetADF / KPSS

github.com/Jrudani21/air-passengers-timeseries ↗

04 / 06

Palmer Penguins
Linear Modeling Field Guide

A comprehensive linear modeling showcase simple linear regression, one-way ANOVA with Tukey post-hoc, two-way ANOVA with interaction, and ANCOVA with adjusted means. R² improves from 0.759 to 0.869 by adding species after flipper length.

ANCOVA adjusted R² = 0.869

RPython ANCOVA ANOVATukey Post-hoc

github.com/Jrudani21/palmer-penguins-linear-models ↗

05 / 06

Medicare Cross-Country
Fraud Detector

Detects providers (NPIs) submitting Medicare claims from a US address and a foreign country simultaneously a documented CMS OIG phantom billing scheme. Five SQL queries (self-join date-overlap detection, risk scoring, country aggregation, specialty benchmarking) plus an interactive Streamlit dashboard with 5 analysis tabs. Built on the CMS Medicare Physician & Other Practitioners PUF schema with 9,976 synthetic claims across 2,000 NPIs.

18 flagged NPIs · $46K+ Medicare payments at risk · HIGH/MEDIUM/LOW risk tiers

SQL Python Streamlit Plotly CTEs Fraud Detection CMS Medicare

github.com/Jrudani21/medicare-fraud-detector ↗

06 / 06

COVID-19 Global
Dashboard

Real-time pandemic statistics application connecting to the disease.sh public API across five interactive views: global KPIs (cases, recoveries, deaths, CFR), 180-day trend analysis, top-15 country rankings, a choropleth map normalised by cases per million, and a searchable data explorer. Data refreshes every 10 minutes via a built-in cache to balance responsiveness with minimal network overhead.

Live data from 200+ countries · sub-second rendering · 180-day historical window

Python Streamlit Plotly pandas REST API Choropleth

github.com/Jrudani21/covid-dashboard ↗

Capabilities

Technical Skills

Languages

R

Python

SQL

Libraries & Frameworks

tidyverse · ggplot2 · forecast

pandas · numpy · matplotlib · seaborn

scikit-learn · statsmodels · pmdarima

Streamlit · Plotly · SQLite

Tools & Workflow

Git & GitHub

R Markdown · Jupyter Notebooks

VS Code

Statistical Methods

Linear & Logistic Regression

One-way / Two-way ANOVA & ANCOVA

Poisson & Negative Binomial GLMs

Time Series (ARIMA / SARIMA)

Hypothesis Testing & Inference

Model Evaluation

AUC / ROC Curves

Confusion Matrix & Classification Report

AIC / BIC Model Comparison

MAPE & Hold-out Validation

Residual Diagnostics & Ljung-Box

Beyond the Data

Interests & Hobbies

PC Building

Designing and assembling custom rigs — from component selection to cable management and overclocking. The hardware side of computing is as satisfying as the software.

Custom white PC build — ASUS RTX GPU and AIO liquid cooling inside White PC case with blue RGB lighting Gaming desk setup with monitor, keyboard, and headset
Custom Rigs Overclocking Cable Mgmt

Circuits & Electronics

Hands-on work with breadboards, microcontrollers, and circuit design. There's something deeply satisfying about making physical things respond to logic.

Electronics desk with breadboard, microcontroller, and multimeter Cloud chamber particle physics experiment with LEDs Breadboard and microcontroller project with colorful jumper wires
Arduino Breadboards Circuit Design

Jazz

From classic Miles Davis to modern fusion — the improvisational structure of jazz feels a lot like exploratory data analysis.

Miles Davis Fusion

Gaming

Strategy, RPGs, and competitive titles. Gaming honed my instinct for systems thinking and pattern recognition long before stats did.

Strategy RPG

Bug Testing

Finding edge cases and breaking things before others do. QA thinking comes naturally — if a model or program can fail, I want to find out how.

QA Edge Cases Debugging

Background

About

I'm Janak Rudani, a statistics student at the University of Manitoba with a focus on applied regression, statistical modelling, and data analysis. My work spans linear models, generalized linear models, and time series all implemented hands-on in both R and Python.

Every project in this portfolio shows the full analytical workflow: data cleaning, exploratory analysis, model selection, assumption checking, and actionable interpretation the same pipeline expected in a professional data analyst role.

Outside of data, I build PCs, tinker with circuits, play and listen to jazz, game competitively, and enjoy breaking software through bug testing. The same curiosity that drives those hobbies drives my work with data.

I'm actively looking for data analyst internships and entry-level roles where I can apply statistical reasoning to real business problems customer behaviour, demand forecasting, A/B testing, or operational analytics.

Full Name

Janak Rudani

Education

University of Manitoba Statistics and Mathematics

Location

Winnipeg, Manitoba, Canada

Primary Languages

R · Python

GitHub

github.com/Jrudani21 ↗

Get In Touch

Let's work together

Open to data analyst internships and entry-level roles. Feel free to reach out.

Janak25rudani@icloud.com