Marketting data analysis
Ngala H.
2024-03-06
Install and load the datarium package
if (!requireNamespace("datarium", quietly = TRUE)) {
install.packages("datarium")
}
library(datarium)
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.5.0 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.1
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorslibrary(caret)## Loading required package: lattice
##
## Attaching package: 'caret'
##
## The following object is masked from 'package:purrr':
##
## lift# Load the marketing data
data(marketing)
head(marketing)## youtube facebook newspaper sales
## 1 276.12 45.36 83.04 26.52
## 2 53.40 47.16 54.12 12.48
## 3 20.64 55.08 83.16 11.16
## 4 181.80 49.56 70.20 22.20
## 5 216.96 12.96 70.08 15.48
## 6 10.44 58.68 90.00 8.64tibble(marketing)## # A tibble: 200 × 4
## youtube facebook newspaper sales
## <dbl> <dbl> <dbl> <dbl>
## 1 276. 45.4 83.0 26.5
## 2 53.4 47.2 54.1 12.5
## 3 20.6 55.1 83.2 11.2
## 4 182. 49.6 70.2 22.2
## 5 217. 13.0 70.1 15.5
## 6 10.4 58.7 90 8.64
## 7 69 39.4 28.2 14.2
## 8 144. 23.5 13.9 15.8
## 9 10.3 2.52 1.2 5.76
## 10 240. 3.12 25.4 12.7
## # ℹ 190 more rowsIntroduction
This document presents an analysis of marketing data.
Methods
EDA and later more rigor methods such as Linear regression will be employed for analysis.
The accuracy of predictions will be measured using the following metrics:
- Mean Absolute Percentage Error (MAPE)
- Root Mean Square Error (RMSE)
EDA
In this section we will inspect the data for; - Bivariate correlation/multi collinearity
Investigate linearity between sales and the explanatory variables through scatterplots
Check for outliers in the explanatory through the use of boxplots etc
Multicollinearity
- Green for values less than 100
- Violet for values between 100 and 200
- Red for values between 200 and 20000
- Dark Red for values above 20000
- Cyan for any other values beyond the specified intervals e.g -negatives
In this section we will conduct a bivariate correlation between all the variables
library(DT)
cov(marketing) %>%
datatable() %>%
formatStyle("facebook",
color = styleInterval(c(100, 200, 20000, Inf), c("green", "violet", "red", "darkred","cyan")))%>%
formatStyle("newspaper",color=(styleInterval(c(100,200,20000,Inf),c('green','violet','red','darkred','cyan'))
))%>%
formatStyle("youtube",color=styleInterval(c(100,200,20000,Inf),c('green','violet','red','darkred','cyan')))library(ggplot2)
# Plot for Facebook
plot_facebook <- ggplot(data = marketing, aes(x = "", y = facebook,fill='facebook')) +
geom_boxplot(show.legend = FALSE) +
labs(y = "Facebook")+
labs(x = "sales")+
scale_fill_manual(values=c('facebook'='yellow'))
# Plot for Youtube
plot_youtube <- ggplot(data = marketing, aes(x = "", y = youtube,fill='Youtube')) +
geom_boxplot(show.legend = FALSE) +
labs(y = "Youtube")+
labs(x = "sales")+
scale_fill_manual(values=c('Youtube'='green'))
# Plot for Newspaper
plot_newspaper <- ggplot(data = marketing, aes(x = "", y = newspaper, fill = "Newspaper")) +
geom_boxplot(show.legend = FALSE) +
labs(y = "Newspaper") +
labs(x = "sales") +
scale_fill_manual(aesthetics = '',values = c("Newspaper" = "cyan"))
# Combine plots using grid.arrange or cowplot
library(gridExtra)##
## Attaching package: 'gridExtra'## The following object is masked from 'package:dplyr':
##
## combinegrid.arrange(plot_facebook, plot_youtube, plot_newspaper, nrow = 1)## Warning: No shared levels found between `names(values)` of the manual scale and the
## data's values.
Conduct supervised learning
model=lm(sales~.,data=marketing)
summary(model)##
## Call:
## lm(formula = sales ~ ., data = marketing)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10.5932 -1.0690 0.2902 1.4272 3.3951
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.526667 0.374290 9.422 <2e-16 ***
## youtube 0.045765 0.001395 32.809 <2e-16 ***
## facebook 0.188530 0.008611 21.893 <2e-16 ***
## newspaper -0.001037 0.005871 -0.177 0.86
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.023 on 196 degrees of freedom
## Multiple R-squared: 0.8972, Adjusted R-squared: 0.8956
## F-statistic: 570.3 on 3 and 196 DF, p-value: < 2.2e-16Splitting data into Train and Testing using caret
index <- createDataPartition(marketing$sales, p = 0.8, list = FALSE)
#all columns and only the rows from the below index
train_data=marketing[index,]
#all columns and only the rows from the below index
test_data=marketing[-index,]Training the data
train_model=lm(sales~.,data=train_data)
summary(train_model)##
## Call:
## lm(formula = sales ~ ., data = train_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10.8224 -1.0227 0.3386 1.4442 3.1212
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.913680 0.426719 9.172 2.41e-16 ***
## youtube 0.045029 0.001632 27.584 < 2e-16 ***
## facebook 0.185754 0.009496 19.561 < 2e-16 ***
## newspaper -0.003462 0.006989 -0.495 0.621
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.062 on 158 degrees of freedom
## Multiple R-squared: 0.8928, Adjusted R-squared: 0.8907
## F-statistic: 438.5 on 3 and 158 DF, p-value: < 2.2e-16conduct prediction
# Generate predictions for test_data
test_data$predicted_sales <- predict(train_model, newdata = test_data)
#Difference between predicted and observed
test_data$difference_observed_predicted=
test_data$predicted_sales-test_data$sales
# Display test_data with predicted sales using datatable()
test_data %>% datatable()MAE=function(x)
{
return(sum(abs(x)) / length(x))
}
MAE(test_data$difference_observed_predicted) %>% tibble()## # A tibble: 1 × 1
## .
## <dbl>
## 1 1.37MAPE <- function(actual, predicted) {
return(mean(abs((actual - predicted) / actual)) * 100)
}
MAPE(test_data$sales,test_data$predicted_sales)%>% tibble()## # A tibble: 1 × 1
## .
## <dbl>
## 1 12.1