Advanced Stadistics in R for Data Analysis

MSc in Data Science Course

Activity 1: Data preprocessing

Goal: Preprocessing of a dataset corresponding to the birth of boys and girls in Spanish hospitals for its subsequent analysis.

1. Loading of the data file and brief description
2. Standardization of qualitative variables
3. Normalization of quantitative variables
4. Missing values
5. Extreme values
6. Summary table of qualitative variables
7. Summary table of quantitative variables
8. Create the clean csv file

Activity 2: Descriptive and inferential analysis

Goal: Apply statistical tools to determine confidence intervals and perform hypothesis tests on data to determine if there is causality in the correlation of data.

1. Descriptive analytics

• Visual descriptive analysis
• Correlation

2. Mean birth weight

• Confidence interval

3. Contrast of the mean value with 3.5kg.

• Hypothesis test

4. Relation to the confidence interval

5. Contrast of mean weight between boys and girls .

• Hypothesis testing
• Normalization of data
• Conclusions

6. Proportion of boys and girls

7. Relationship between underweight and mother smoking

• Causality analysis of underweight with the smoking mother factor. It is concluded that smoking does significantly affect low birth weight.

8. Analysis of low birth weight cases (with gestation >36).

Activity 3: Predictive modeling

Goal: Building statistical models to obtain accurate and meaningful information from the data

1. Linear regression model

• Univariate linear regression model
• Multiple linear regression model (quantitative regressors) .

2. Model diagnosis

3. Model prediction

4. Logistic regression model

• OR (Odds Ratio) estimation .
• Logistic regression model
• Prediction
• Fitting
• ROC curve

Activity 4: Analysis of variance

Goal: Analysis of variance of data, treatment of outliers and inferential statistics.

1. Descriptive statistics and visualization

2. Inferential statistics

• Confidence interval
• Comparison test of two means
• Null and alternative hypothesis
• Choice of analysis method
• Statistical calculations of contrast, critical value and p-value with a confidence interval of 95%.

3. Non-parametric contrast

4. Regression

5. One-factor analysis of variance (ANOVA)

• Null and alternative hypothesis
• Graphical display
• Normality of residuals
• Homoscedasticity of the residuals

6. Nonparametric ANOVA