Skip to content
Analysis of Website Redesign
  • AI Chat
  • Code
  • Report
  • This code loads necessary packages, supplies the data as a dataframe (df), and shows the first six rows as a table.

    install.packages("naniar")
    install.packages("tidyverse")
    library(tidyverse)
    library(naniar)
    df <- readr::read_csv('./data/redesign.csv')
    head(df)
    1. Analyze the conversion rates for each of the four groups: the new/old design of the landing page and the new/old pictures.

    This code checks for any missing data.

    miss_var_table(df)

    There are no missing values in the dataframe. The following code calculates the conversion rates for each of the four groups and displays these rates in a table.

    
    df %>%
      group_by(treatment, new_images) %>%
     summarize(proportion_converted = mean(converted))
      
    1. Can the increases observed be explained by randomness? (Hint: Think A/B test)

    The following code tests for whether the conversion rates displayed above are statistically different from one another.

    df$treatment <- factor(df$treatment, levels = c("no", "yes"))
    df$new_images <- factor(df$new_images, levels = c("no", "yes"))
    model <- glm(converted ~ treatment * new_images, data = df, family = "binomial")
    summary(model)

    This analysis reveals that the new landing page increases conversion rate regardless of whether the new images are also added. Additionally, there is no statistically significant effect of the new images on conversion rates. In order to convert the logistic coefficient to a more intuitive value to interpret, I converted the treatment coefficient to an odds ratio, which indicates that there is a 13.73% greater relative chance of converting when exposed to the new landing page relative to the old landing page.

    exp(.12869)
    1. Which version of the website should they use?

    I recommend retaining the old images and utilizing the new landing page.