1. Introduction:

This project analyzes team performance data from a dataset named “e1_data.csv”. The dataset contains information about various teams, their assessment scores, and the locations where the assessments were conducted. My analysis aims to provide insights into team performance, score distributions, and geographical patterns.

2. Dataset Overview:

The “e1_data” dataset consists of 91 rows and 5 columns:

• teamno: Team identifier
• num_people: Number of people in each team
• assessment_date: Date of the assessment
• score: Performance score
• location: Place where the assessment was conducted

For example, the first row of data shows: Team 1, with 8 members, scored 96.2 in an assessment conducted on May 15th in Norfork.

3. Methodology

I use R for my analysis, leveraging packages such as tidyverse for data manipulation and ggplot2 for visualization. My approach involves data processing and data visualizations (bar charts, histograms, density plots, line graphs, geographical mapping of scores).

4. Expected outcome

This analysis will provide insights into:

• Top-performing teams as well as low-performing teams
• Distribution of scores and any patterns or anomalies
• Geographical variations in scores across Massachusetts counties

Part 1: Bars and Histograms

For the first part of this project, I am going to focus on team-by-team performance.

For part 1a., I take the data and create a subsidiary data set which has one row for each team and then the average (mean), maximum, and minimum scores that they achieved.

e1_df <- read_csv("/Users/linh/PROJECTS/R/e1_data.csv") 

## Rows: 90 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (2): assessment_date, location
## dbl (3): teamno, num_people, score

e1_stat <- e1_df %>%
  group_by(teamno) %>%
  summarize(mean_score = mean(score),
            max_score = max(score),
            min_score = min(score))
e1_stat <- e1_stat[order(e1_stat$mean_score, decreasing = TRUE),]
e1_stat

## # A tibble: 10 × 4
##    teamno mean_score max_score min_score
##     <dbl>      <dbl>     <dbl>     <dbl>
##  1      3       94.5      96.2      93.1
##  2      8       94.5      95.9      93.2
##  3      4       93.2      95.3      90.7
##  4      5       92.6      93.9      91.1
##  5      2       92.4      94.5      89.9
##  6      6       92.1      94.2      90.2
##  7     10       91.8      91.8      91.8
##  8      1       91.8      96.2      88.9
##  9      7       91.8      94.8      89.7
## 10      9       89.3      91.3      87.8

Based on the table e1_stat:

• Team 3 consistently outperformed others with the highest mean score (94.54) and maximum score (96.2).
• Most teams maintained relatively consistent performance, with mean scores ranging from 89.31 to 94.54.
• Team 1 showed the widest performance range, with scores varying from 88.9 to 96.2.
• The lowest minimum score (87.8) belonged to Team 9, indicating potential areas for improvement.

For part 1b, I create a bar chart with each team number and the number of people who are on that team, then color in the fill according to their average scores.

summarized <- e1_df %>% group_by(teamno, num_people) %>% summarize(avg_score = mean(score)) 

summarized %>% ggplot(aes(x = reorder(factor(teamno), avg_score), y = num_people, fill = avg_score)) + 
  geom_col() +
  geom_text(aes(label = num_people), hjust = -0.2, color = "black", size = 3.0) +
  #scale_fill_gradient(low = 'lightblue', high = 'darkblue')
  scale_fill_distiller(palette = "RdYlBu", direction = 1) + 
  labs(title = 'Relationship Between Team Size and Average Scores', 
       x = 'Team Number',
       y = 'Team Size',
       fill = 'Average Score') +
  theme_minimal() + coord_flip() +
   ylim(0, max(summarized$num_people) * 1.1)

Based on the bar chart, we can draw the conclusion that larger team size does not necessarily correlate with better performance:

• Top Performers: Teams 3 and 8, both with 10 members, achieved the highest average scores (around 94). This suggests that a team size of 10 might be optimal for high performance in this context.
• Lowest Performance: Team 9, with 11 members, shows the lowest average score (around 90).
• The majority of teams fall in the middle range of performance (scores between 91-93), regardless of their size.

For part 1c, I create a histogram and a density plot, separately, showing the distribution of average scores over the time period.

summarized_date <- e1_df %>% group_by(assessment_date) %>% summarise(avg_score = mean(score))
                      
summarized_date %>% ggplot(aes(x = avg_score)) +
  geom_histogram(binwidth = 0.5, col = "black", fill = "blue") + 
  labs(title = 'Distribution of Average Scores',
  x = 'Average Score',
  y = 'Frequency')

summarized_date %>% ggplot(aes(x = avg_score)) +
  geom_density(color = "black", fill = "orange") + 
  labs(title = 'Distribution of Average Scores',
  x = 'Average Score',
  y = 'Density')

For part 1d, I overlay both the histogram and the density plot into the same chart.

summarized_date %>% ggplot(mapping = aes(x = avg_score)) +
  geom_histogram(aes(y = after_stat(density), binwidth = 0.5)) +
  geom_density(kernel = "epanechnikov", color = "orange") +
  labs(
    title = "Distribution of Average Scores",
    x = "Average Score",
    y = "Density"
  ) 

## Warning in geom_histogram(aes(y = after_stat(density), binwidth = 0.5)):
## Ignoring unknown aesthetics: binwidth

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Based on the density plot:

• The distribution appears to be roughly normal, with a slight right skew. The peak of the density curve suggests that the most common average scores are around 92-93.
• High Performers: There are a few teams with scores above 94, representing the top performers. These teams might have best practices that could be valuable to study and potentially implement across other teams.
• Low Performers: The left tail of the distribution shows a few teams with scores below 91. These teams may require additional support to improve their performance.

Part 2

For part 2, I am interested in the time and place aspects of the scoring.

For Part 2a, I create a new (temporary) data frame to support the visualization which shows the average performance of all teams in aggregate month-by-month. Then I create a second chart which shows a separate line for each team; use the “linetype” aes to differentiate.

temp_df <- e1_df %>%
  mutate(assessment_date = as.Date(assessment_date, format = "%B %d")) %>%
  group_by(teamno, month = month(assessment_date)) %>%
  summarize(avg_score = mean(score))

temp_df %>% ggplot(aes(x = month, y = avg_score, group = teamno, color = factor(teamno))) +
  geom_line() +
  labs(title = "Trend of Team Performance (May-Sep)",
       x = "Month",
       y = "Average Score",
       color = "Team Number")

Based on the line chart, there is a general downward trend in performance for most teams over the five-month period, suggesting systemic factors affecting team productivity or scoring.

Finally, for part 2b, I create a geographical plot of average scores by county.

map_score <- e1_df %>% group_by(location) %>% summarize(avg_score = mean(score)) %>% mutate(location = tolower(location))

state_map <- map_data('county', 'mass') %>% rename(location = subregion) %>% left_join(map_score, by = 'location')

p <- ggplot(state_map, aes(long, lat, group=group))

p +  geom_polygon(aes(fill = avg_score), color = 'yellow') +
  coord_map(projection ='albers', lat0 = 39, lat1 = 45) + 
  scale_fill_gradientn(colors = c("#4891FF", "#4183E7", "#3A75CF", "#3367B7", "#2C599F", "#254B87", "#1E3D6F", "#172F57")) +
  labs(title = "Average Scores by County in Massachusetts",
      x = "Longitude",
      y = "Latitude",
      fill = "Average Score"
  )

Based on the map:

• There is a noticeable east-west divide in performance.Eastern counties, particularly those in and around the Greater Boston area, show higher average scores (darker blue). Western counties generally display lower average scores (lighter blue).

• The highest-performing counties (darkest blue) are concentrated in the eastern part of the state, likely including Suffolk County (Boston) and its surrounding areas. This could indicate advantages in urban and suburban areas, possibly due to greater access to resources, educational institutions, or economic opportunities.

• Coastal counties show some variation in performance. Cape Cod and the islands (likely Nantucket and Martha’s Vineyard) appear to have moderately high scores.