Ranalytics

This repository contains the weekly work exercises for data science.

by Carlos Utrilla Guerrero.

You can find totally available source here:

Practical cases:

- Biomakers:

Do patients that express one or more of these biomarkers have better, or worse, outcomes that those that do not express these biomarkers?

- Forecasting credit default probability:

On this study, we sought to explore some way to forecast the chance that a specific customer, who want to apply for a specific bank product, will default on his/her payments. That said, trying to estimate the probability of default. Give some application to estimate the credit-worthiness of a customer.

- Financial Data:

We analyse three major stocks indices, Dow Jones, DAX and Nikkei in relation to the exchange rate between US Dolar, Euro and Yen. Research Questions: how large was the impact of the recent financial crisis on the stock markets?; how correlated is the development between the stock market?; what is the relationship between stock market returns and exchange rates?.

- Robust Statistics:

A practical case to explore issues around the topics of regression assumption, non parametric tests, and bootstrap analyses.

- New Statistics:

A use-case that show issues surrounding power analysis, confiance intervals and bayesian statistics.

- Machine Learning:

A set of practical examples that dive into the ML tasks and implementation like feature extraction, modeling and fitting models, optimization and prediction.

- Analyzing Sales data:

In this case study we perform data wrangling and merge the two datasets merged in one. It also goes beyond and addresses research questions such as:

• how large the fluctuation of sales were per store.
  
• whether sales number go up or down in holidays.
  
• how well we can predict sales numbers from other variables, time series analysis predict future turnovers.
  Because there are many timepoints per store, we aggregate the values to interpret p-values using mixed effect models.
  

- Time Series Analysis:

Quarterly beer sales data have been deeply explored. It uses Holt-Winters seasonal method to capture seasonality and predict for next two years. Also uses ARIMA model to predict next 2 years.