GlobalHealth_stats.Rmd

---
title: "Global Health Stats"
output: 
  html_notebook: 
    toc: yes
    theme: spacelab
---



<style type="text/css">
a {color: #3d0066; font-size: 11;}
p {color: #000000; font-family:"monaco" font-size:13;}
h1 {color: #3d0066 }
h2 {color: #6b00b3 }
h3 {color: #9900ff }
h4 {color: #cc0099 }
ul {list-style-type: circle; color: #6b00b3 }
ol {color: #cc0099; font-size:11; }
p1 {color: #004d4d;}
li {color: #004d4d;} 

</style>

<t>
![](RStudio-Logo 1.png)</t>

# Install Gapminder library
```{r Install Gapminder Library}
install.packages("gapminder")
library(gapminder)
data(gapminder)
```

# Get the Summary Statistics
```{r Get Summary Stats}
summary(gapminder)
```
# Get the mean of a column
```{r Mean}
# the mean of the GDP per Capita
gdp.mean = mean(gapminder$gdpPercap)
gdp.mean
```

## use the attach function to get column values
```{r}
attach(gapminder)
median(pop)
```

## quick histogram 
```{r}
# histogram of life expectancy log base
log.lifeExp = log(pop)
hist(log.lifeExp)

```
## quick boxplot
```{r}
# life expectancy based on continent
boxplot(lifeExp ~ continent)
```
## quick scatterplot
```{r}
# life expectancy (dep var) on the Y axis
# gdpPercap (indep var) on X axis
plot(lifeExp ~ log(gdpPercap))
```

# install the dplyr library
The pipe function %>% with the select function for the column variables of interest. The second pipe is like the "and then" with the filter function to select row variable based on a condition.
The mutate function is used to create a new variable from a data set. In order to use the function, we need to install the dplyr package.

data %>% select(col.1, col.2, col.3) %>% 
  filter(col.1 =="Type 2")  %>% 
  mutate(BMI = weight/ (height^2))  %>% 
  summarize(Average_BMI = mean(BMI))

```{r}
library(dplyr)

gapminder %>% 
  select(country, lifeExp)
```


```{r}
gapminder %>% 
  select(country, lifeExp) %>% 
  filter(country =="Norway" | country =="Canada") %>% 
  group_by(country) %>% 
  summarise(Avg_LifeExp = mean(lifeExp))
```

## Variation of sample
Differences by chance/ no difference (Null hypothesis) which t-test is used with a p-value p= 0.05. Null is p < 0.05. This is compared to real change (alternative hypothesis) p > 0.05. The 95% confidence interval.

```{r}

df.1 = gapminder %>% 
  select(country, lifeExp) %>% 
  filter(country =="Norway" | country =="South Africa") 

# use the t-test to determine statistical differance
t.test(data= df.1, 
       lifeExp ~ country)

```

The odds of our sample difference is the p-value, so for this case we reject the null hypothesis based on the differences of means is NOT zero. 


```{r}
library(ggplot2)

gapminder %>% 
  filter(gdpPercap < 50000) %>% 
  ggplot(aes(x= log(gdpPercap), 
             y= lifeExp,
             col= continent,
             size= pop)) +
  geom_point(alpha= 0.3) +
  geom_smooth(method = lm)

```

```{r}
# linear regression
summary(lm(lifeExp ~ gdpPercap))
```


```{r}
# multiple linear regression
summary(lm(lifeExp ~ gdpPercap + pop))

# the P-value is the Pr
```

The p-values are 2e-16 and 4.72e-05


Population => sample : measure variables
Categorical = groups, summarize
Numerical = summarize : range, IQR, mean, median

sample tests:

- 1 sample proportion test
- chi-squared test
- t-test
- ANOVA
- correlation test 

1 categorical (binary) : 1 sample proportion test <br>
2 categorical (child/adult/senior) : chi squared <br>
1 numeric : t-test <br>
1 numeric AND 1 categorical : t-test or ANOVA <br>
2 numeric : correlation <br>


## Null and Alt hypothesis

- is there a difference? what are chances/ probability of difference?
- Null: no difference
- ALt: there is no difference
- alpha value = 0.05
- test: 1 sample proportion test, get p-value
- if p-value < alpha = reject null, statistical significant

"is the average height different significant between groups?"

- Null: there is no difference
- Alt: there is a difference
- alpha= 0.05
- test: t-test, compare averages, p-value
- test: ANOVA/ correlation, compare weight vs height
- if p-value < alpha = reject null

correlation -1 to +1