WHO ARE THE STAKEHOLDERS?

ABOUT THE DATA

WHERE IS YOUR DATA STORED?

1. • The dataset can be accessed via Kaggle, a website that publishes datasets.

HOW IS THE DATA ORGANIZED?

1. • The dataset contains both long (narrow) and wide versions. However, some .csv files are only stored in a long version, thus all data will be utilized in a long format.

HOW ARE YOU ADDRESSING LICENSING PRIVACY, SECURITY, AND ACCESSIBILITY?

1. • The dataset is shared by Mobius, the collaborator, with a CC0 Public domain license. Thus, anyone can access the same dataset.

ARE THERE ANY PROBLEMS OR ISSUES WITH BIAS OR CREDIBILITY IN THIS DATA?

1. • We will use ROCCC (Reliable, Original, Comprehensive, Current, & Cited) to determine if it is a good data source or not

Reliable: The data is rated LOW on this criteria since 30 respondents are not enough samples for an accurate analysis.

Original: It is rated LOW because it came from the third-party provider (Amazon Mechanical Turk)

Comprehensive: This is rated MEDIUM on this criteria as the data provides the data we might need for the analysis. However, it lacks demographic data which can be useful for analysis.

Current: The data is collected 7 years ago, thus this is rated LOW on this criteria.

Cited: The data is collected from a third-party provider, hence it is rated LOW.

1. • Overall, it is not a recommended data source to begin with. 

HOW DOES IT HELP YOU ANSWER YOUR QUESTION?

• Although the data source is outdated and can be from a third-party provider, the data provides sufficient information that we need to do the business task.

THE TOOLS THAT I USED ARE

• MS Excel to inspect and rename the files 

• R to clean, analyze, and visualize the datasets

STEPS IN THE PROCESS DATA

First, I downloaded the zip file from the Kaggle website to get the .csv files. For this project, I used the narrow files.

Using MS Excel, I renamed the narrow files into simpler names.

• dailyActivity_merged.csv -> Daily_Activity.csv

• dailyCalories_merged.csv -> Daily_Calories

• dailyIntensities_merged.csv -> Daily_Intensities

• dailySteps_merged.csv -> Daily_Steps

• heartrate_seconds_merged.csv -> Heartrate_Seconds.csv

• hourlyCalories_merged.csv -> Hourly_Calories.csv

• hourlyIntensities_merged.csv -> Hourly_Intensities.csv

• hourlySteps_merged.csv -> Hourly_Steps.csv

• minuteCaloriesNarrow_merged.csv -> Minutes_Calories.csv

• minuteIntensitiesNarrow_merged.csv -> Minutes_Intensities

• minuteMETsNarrow_merged.csv -> Minutes_METs

• minuteSleep_merged.csv -> Minutes_Sleep.csv

• minuteStepsNarrow_merged.csv -> Minutes_Steps.csv

• sleepDay_merged.csv -> Sleep_Day.csv

• weightLogInfo_merged.csv -> WeightLogInfo.csv

After renaming and reformatting the data, I installed and imported the datasets needed for the analysis using R

• Installing packages

install.packages(“tidyverse”)

library(tidyverse)

library(readr)

library(ggplot2)

library(dplyr)

• Importing datasets

Daily_Activity <- read_csv("Daily_Activity.csv")

Daily_Calories <- read_csv("Daily_Calories.csv")

Daily_Intensities <- read_csv("Daily_Intensities.csv")

Daily_Steps <- read_csv("Daily_Steps.csv")

Heartrate_Seconds <- read_csv("Heartrate_Seconds.csv")

Hourly_Calories <- read_csv("Hourly_Calories.csv")

Hourly_Intensities <- read_csv("Hourly_Intensities.csv")

Hourly_Steps <- read_csv("Hourly_Steps.csv")

Minutes_Calories <- read_csv("Minute_Calories.csv")

Minutes_Intensities <- read_csv("Minutes_Intensities.csv")

Minutes_METs <- read_csv("Minutes_METs.csv")

Minutes_Sleep <- read_csv("Minutes_Sleep.csv")

Minutes_Steps <- read_csv("Minutes_Steps.csv")

Sleep_Day <- read_csv("Sleep_Day.csv")

Weight_Log_Info <- read_csv("WeightLogInfo.csv")

DATA MANIPULATION AND CLEANING

Inspecting the data

• sapply(list(Daily_Activity, Daily_Calories, Daily_Intensities, Daily_Steps, Heartrate_Seconds, Hourly_Calories, Hourly_Intensities, Hourly_Steps, Minutes_Calories, Minutes_Intensities, Minutes_METs, Minutes_Sleep, Minutes_Steps, Sleep_Day, Weight_Log_Info), head)

• sapply(list(Daily_Activity, Daily_Calories, Daily_Intensities, Daily_Steps, Heartrate_Seconds, Hourly_Calories, Hourly_Intensities, Hourly_Steps, Minutes_Calories, Minutes_Intensities, Minutes_METs, Minutes_Sleep, Minutes_Steps, Sleep_Day, Weight_Log_Info), str)

• sapply(list(Daily_Activity, Daily_Calories, Daily_Intensities, Daily_Steps, Heartrate_Seconds, Hourly_Calories, Hourly_Intensities, Hourly_Steps, Minutes_Calories, Minutes_Intensities, Minutes_METs, Minutes_Sleep, Minutes_Steps, Sleep_Day, Weight_Log_Info), colnames)

• sapply(list(Daily_Activity, Daily_Calories, Daily_Intensities, Daily_Steps, Heartrate_Seconds, Hourly_Calories, Hourly_Intensities, Hourly_Steps, Minutes_Calories, Minutes_Intensities, Minutes_METs, Minutes_Sleep, Minutes_Steps, Sleep_Day, Weight_Log_Info), glimpse)

Cleaning the data

• sum(is.na(Daily_Activity))

[1] 0

• sum(is.na(Daily_Calories))

[1] 0

• sum(is.na(Daily_Intensities))

[1] 0

• sum(is.na(Daily_Steps))

[1] 0

• sum(is.na(Heartrate_Seconds))

[1] 0

• sum(is.na(Hourly_Calories))

[1] 0

• sum(is.na(Hourly_Intensities))

[1] 0

• sum(is.na(Hourly_Steps))

[1] 0

• sum(is.na(Minutes_Calories))

[1] 0

• sum(is.na(Minutes_Intensities))

[1] 0

• sum(is.na(Minutes_METs))

[1] 0

• sum(is.na(Minutes_Sleep))

[1] 0

• sum(is.na(Minutes_Steps))

[1] 0

• sum(is.na(Sleep_Day))

[1] 0

• sum(is.na(Weight_Log_Info))

[1] 65

• View(Weight_Log_Info)

While checking if there are NA values, I saw that Weight_Log_Info has 65 NA values in the fat column. However, there were 2 cells that have data in the said column, thus, I removed the column

• Weight_Log_Info <- subset(Weight_Log_Info, select = -Fat)

• sum(is.na(Weight_Log_Info))

  [1] 0

I checked if there are NULL values in each dataset

• sum(is.null(Daily_Activity))

  [1] 0

• sum(is.null(Daily_Calories))

  [1] 0

• sum(is.null(Daily_Intensities))

  [1] 0

• sum(is.null(Daily_Steps))

  [1] 0

• sum(is.null(Heartrate_Seconds))

  [1] 0

• sum(is.null(Hourly_Calories))

  [1] 0

• sum(is.null(Hourly_Intensities))

  [1] 0

• sum(is.null(Hourly_Steps))

  [1] 0

• sum(is.null(Minutes_Calories))

  [1] 0

• sum(is.null(Minutes_Intensities))

  [1] 0

• sum(is.null(Minutes_METs))

  [1] 0

• sum(is.null(Minutes_Sleep))

  [1] 0

• sum(is.null(Minutes_Steps))

  [1] 0

• sum(is.null(Sleep_Day))

  [1] 0

• sum(is.null(Weight_Log_Info))

  [1] 0

There were no nulls in the datasets. After looking if the datasets have NAs or nulls, I counted the unique values using distinct

• Daily_Activity %>% distinct(Id)

• Daily_Calories %>% distinct(Id)

• Daily_Intensities %>% distinct(Id)

• Daily_Steps %>% distinct(Id)

• Heartrate_Seconds %>% distinct(Id)

• Hourly_Calories %>% distinct(Id)

• Hourly_Intensities %>% distinct(Id)

• Hourly_Steps %>% distinct(Id)

• Minutes_Calories %>% distinct(Id)

• Minutes_Intensities %>% distinct(Id)

• Minutes_METs %>% distinct(Id)

• Minutes_Sleep %>% distinct(Id)

• Minutes_Steps %>% distinct(Id)

• Sleep_Day %>% distinct(Id)

• Weight_Log_Info %>% distinct(Id)

Next, I counted the number of observations in each dataframe

• nrow(Daily_Activity)

  [1] 940

• nrow(Daily_Calories)

  [1] 940

• nrow(Daily_Intensities)

  [1] 940

• nrow(Daily_Steps)

  [1] 940

• nrow(Heartrate_Seconds)

  [1] 683001

• nrow(Hourly_Calories)

  [1] 22099

• nrow(Hourly_Intensities)

  [1] 22099

• nrow(Hourly_Steps)

  [1] 22099

• nrow(Minutes_Calories)

  [1] 1048575

• nrow(Minutes_Intensities)

  [1] 1048575

• nrow(Minutes_METs)

  [1] 1048575

• nrow(Minutes_Sleep)

  [1] 188521

• nrow(Minutes_Steps)

  [1] 1048575

• nrow(Sleep_Day)

  [1] 413

• nrow(Weight_Log_Info)

  [1] 67

Then, I merged the datasets with the same set of Id number and rows using inner joins

• inner_join(Daily_Activity, Daily_Calories, by = "Id")

• inner_join(Daily_Activity, Daily_Intensities, by = "Id")

• inner_join(Daily_Activity, Daily_Steps, by = "Id")

• inner_join(Minutes_Calories, Minutes_Intensities, by = "Id")

Note: Unfortunately, R crashed while running the formula for other datasets. Hence, I used MS Excel to merge the other datasets

• Hourly_Calories, Hourly_Intensities, & Hourly_Steps -> Hourly_Cal_Int_Steps

• Minutes_Calories, Minutes_Intensities, Minutes_METs, & Minutes_Steps -> Minutes_Cal_Int_METs_Steps

I uploaded the merged datasets in R and double-checked again if there are NAs or NULLs 

• Hourly_Data_Merged <- read_csv("Hourly_Cal_Int_Steps.csv")

• Minutes_Data_Merged_ExceptSleep <- read_csv("Minutes_Cal_Int_METs_Steps.csv")

• sum(is.na(Hourly_Data_Merged))

  [1] 0

• sum(is.na(Minutes_Data_Merged_ExceptSleep))

  [1] 0

• sum(is.null(Hourly_Data_Merged))

  [1] 0

• sum(is.null(Minutes_Data_Merged_ExceptSleep))

  [1] 0

I wanted to get the descriptive statistics by performing the summary function.

DATA FINDINGS

1. Average steps per day: 7638 steps

2. Average total distance walked per day: 5.9 km

3. Average sedentary hours: 16 ½ hours (991.2 minutes)

4. Average calories they burned: 2306 calories

5. Average heartrate: 79 beats per minute

6. Average total time of sleep per day: approximately 7 hours or 419.5 minutes

7. Calculate the number of minutes awake in bed

   • Sleep_Day <- Sleep_Day %>% 

mutate(c=TotalTimeInBed-TotalMinutesAsleep)

7. • Sleep_Day <- Sleep_Day %>%

rename("MinutesAwakeInBed" = "Difference")

8. Average minutes awake in bed: 39 minutes

9. Average weight of users: 72kg or 158.8 pounds

10. Average BMI: 25.19 (Overweight)

11. Correlation of very Active Minutes & Calories

    • cor(Daily_Activity$VeryActiveMinutes, Daily_Activity$Calories, method = "spearman")

    • [1] 0.5395405

In order to know the number of application usage in the whole week, let’s transform dates into days

• Library(lubridate)

• Daily_Activity$New_Date <- mdy(Daily_Activity$ActivityDate)

• Daily_Activity$weekday <- weekdays(Daily_Activity$New_Date)

Additionally, I made a bar chart to visualize the other important details.

LET'S ANSWER THE BUSINESS TASKS

WHAT ARE SOME TRENDS IN SMART DEVICE USAGE?

1. • 991.2 minutes or 16.5 hours is the average recorded sedentary minutes among participants. This is quite alarming since 7 hours of sedentary behavior can place people at great risk of developing cardiovascular diseases and cancer

   • Participants have an average BMI of 25.19, which is classified as “Overweight”.

   • According to a recent study, people who walk at least 7,000 steps a day can help lower the risk of early death. Thus, this will be a great start for those who want to be healthy.

   • The average sleep time is 7 hours per day, which is recommended for adults.

   • When it comes to Bellabeat usage, participants choose to use the application during weekdays (except for Monday). This might indicate the preference of individuals to be active after work, which is demonstrated in hourly steps.

   • Expectedly, there is a positive correlation between the total number of steps and calories, as well as very active minutes and calories. This indicates the importance of physical activity in order to stay in shape.

HOW COULD THESE TRENDS APPLY TO BELLABEAT CUSTOMERS?

1. • These trends will help the customers in an indirect way. By presenting these findings to the stakeholders, they might do action to improve the smart devices which aim to promote and encourage healthy behaviors to the customers.

HOW COULD THESE TRENDS HELP BELLABEAT MARKETING STRATEGY?

1. • There are actions that can be taken to improve Bellabeat’s marketing strategy

TAKE NOTE THAT THERE ARE LIMITATIONS TO THE DATA

• There is no demographic data that was taken. They should include the demographics of the participants, such as age, gender, ethnicity,etc. This will help the analysts to gain more insights to the company for further actions.

• Having 30 participants does not truly reflect the findings provided. The number of participants should represent each category of individuals and increase the number of participants.

• The data is outdated and it needs to be updated each year to gain accurate reports.