Chess Breakdown

devtools::install_github("JaseZiv/chessR")

library(chessR)

library(dplyr)

library(ggplot2)

library(stringr)

library(lubridate)

```

# Pulling my data with the chessR package, username: rainybeater123

```{r}

rainybeater <- get_raw_chessdotcom(usernames = "rainybeater123")

rainybeater <- rainybeater %>%

  mutate(

    result = case_when(

      White == "rainybeater123" & Result == "1-0" ~ "Win",

      Black == "rainybeater123" & Result == "0-1" ~ "Win",

      Result == "1/2-1/2" ~ "Draw",

      TRUE ~ "Loss"

    ))

```

## Looking at total win rate and then breaking it down by ECO (openings)

```{r}

summary <- rainybeater %>%

  group_by(result) %>%

  summarise(count = n()) %>%

  mutate(win_rate = count / sum(count) * 100)

ecowin <- rainybeater %>%

  group_by(ECO) %>%

  summarise(

    total_games = n(),

    wins = sum(result == "Win"),

    draws = sum(result == "Draw"),

    losses = sum(result == "Loss"),

    win_rate = (wins / total_games) * 100

  ) %>%

  arrange(desc(win_rate))

head(ecowin)

```

# Joining the top and bottom 10 eco positions in the same dataframe

* I set the minimum number of games to 10 for each opening

```{r}

top_eco <- ecowin %>%

  filter(total_games > 10) %>%  

  slice_max(win_rate, n = 10) 

bot_eco <- ecowin %>% 

  filter(total_games > 10) %>% 

  slice_min(win_rate, n = 10)

topbot_eco <- rbind.data.frame(top_eco, bot_eco)

topbot_eco$top_or_bot <- 0

topbot_eco_1 <- topbot_eco %>% 

  mutate(top_or_bot = ifelse(win_rate > 50, "Win", "Loss"))

print(topbot_eco_1)

```

 * Clearly, there are some openings that I should avoid and others that I tend to do better with

```{r, echo = FALSE}

ggplot(topbot_eco_1, aes(x = ECO, y = win_rate, fill = top_or_bot)) +

  geom_bar(stat = "identity") +

  labs(

    title = paste("Top and Bottom Openings (ECO) for rainybeater123"),

    x = "ECO Code",

    y = "Win Rate (%)"

  ) +

  facet_wrap(~ top_or_bot, scales = "free_x") +

  theme_bw() +

  theme(legend.position = "none")

```

# Setting up some new variables

 * Determining if I went first or second

 * Assigning my ELO score to each game

 * Game duration

```{r}

first_second <- rainybeater %>% 

  mutate(first_second = ifelse(White %in% "rainybeater123", "first", "second"), 

         my_color = ifelse(White %in% "rainybeater123", "white", "black"), 

         my_elo = as.numeric(ifelse(my_color == "white", WhiteElo, BlackElo)), 

         Date = as.Date(Date), 

         StartTime = hms(StartTime),

         EndTime = hms(EndTime),

         Duration = as.numeric(EndTime - StartTime, units = "minutes")) %>% 

  filter(TimeClass %in% "rapid") %>% 

  filter(Duration > 0)

ecocount <- first_second %>%

  group_by(result, ECO) %>%

  mutate(ECO_count = n())

```

# My ELO Scoring Over Time

 * I was on a pretty steady increase from March '21 to March '22

 * I stayed pretty stagnant for the next year or so...

 * Let's forget about March '23 to March '24!

 * I have been getting back on track recently...

```{r, echo = FALSE}

ggplot(first_second) +

  geom_line(aes(x = Date, y = my_elo)) +

  labs(

    title = "My Chess Elo Since 2021",

    subtitle = "Split by 6 month increments, March 2021 - October 2024", 

    x = "Date",

    y = "ELO"

  ) +

  theme(

    legend.position = "none"

  ) +

  theme_bw() +

  scale_x_date(date_breaks = "6 month", date_labels = "%m-%y")

```

# Barplots for wins and lossess by type of ending

```{r, echo = FALSE}

wintype <- first_second %>% 

  filter(result == "Win") %>% 

  mutate(wonhow = str_extract(Termination, "won by checkmate|won by resignation|won on time")) %>% 

  filter(!is.na(wonhow)) %>% 

  mutate(wonhow = as.factor(wonhow))

ggplot(wintype) +

  geom_bar(aes(x = wonhow, fill = wonhow), position = "dodge") +

  labs(

    title = "Distribution of Chess Wins",

    subtitle = "For all games that were not aborted", 

    x = "How I Won",

    y = "Count",

    fill = "How I Won"

  ) +

  theme_bw() +

  theme(

    legend.position = "top"

  )

losstype <- first_second %>% 

  filter(result == "Loss") %>% 

  mutate(losshow = str_extract(Termination, "won by checkmate|won by resignation|won on time")) %>% 

  filter(!is.na(losshow)) %>% 

  mutate(losshow = as.factor(losshow)) %>% 

  mutate(losshow = str_replace(losshow, "won by checkmate", "lost by checkmate"),

         losshow = str_replace(losshow, "won by resignation", "lost by resignation"),  

         losshow = str_replace(losshow, "won on time", "lost on time"))

ggplot(losstype) +

  geom_bar(aes(x = losshow, fill = losshow), position = "dodge") +

  labs(

    title = "Distribution of Chess Losses",

    subtitle = "For all games that were not aborted. I have always been too stubborn to resign.", 

    x = "How I Lost",

    y = "Count",

    fill = "How I Lost"

  ) +

  theme_bw() +

  theme(

    legend.position = "top"

  )

```

# This is a distribution of results based on how long the game goes on

* The longer the games go, the more likely a draw becomes

* Some quick wins are usually due to opponent resignations

```{r, echo = FALSE}

ggplot(first_second) +

  geom_density(aes(x = Duration, fill = result), alpha = .5) +

    scale_fill_manual(values = c("Win" = "gold", "Loss" = "red", "Draw" = "blue")) +   labs(

    title = "Distribution of Time for Wins/Losses/Draws", 

    x = "Duration of games in minutes", 

    y = "Frequency"

  ) +

  theme_bw() 

# +  facet_wrap(~result)