Electra_Grid Activity

DISEÑO DE EXPERIMETOS

DISEÑO DE EXPERIMENTOS 1

Client Memo: Capital Allocation Using MPT

Tesla, Toyota, GM - Part 2 MPT Assignment

TextPredictor is a Shiny web app that predicts the next word in a user-entered phrase using unigram, bigram, and trigram frequency tables with a backoff strategy. The system is fast, lightweight, and demonstrates practical applications of R, Shiny, and NLP for text prediction. A 5-slide RPubs presentation explains the model, data, and app functionality.

This project evaluates the financial impact of a workplace wellness programme using data from a randomised control trial (RCT). The analysis begins by cleaning the dataset and renaming variables for clarity. It then explores the data through summary statistics and visualisations, highlighting a potential selection bias where healthier employees are more likely to enrol. To address this, the study contrasts a standard OLS regression, which misleadingly suggests cost savings due to the "healthy user" effect, with an instrumental variables (IV) regression. The IV analysis uses the random assignment of the programme as an instrument to isolate the true causal effect, ultimately revealing that the wellness programme has no statistically significant impact on reducing employee medical spending.

I use simulations to show how the spread of the sample distribution changes with sample size and show that this produces the standard error of the mean.