NaiveStudentGradePredict.Rmd

---
title: "Predict Student's Final Grade"
output:
  html_document:
    df_print: paged
---

<p>

Created by Jonathan Calderon Chavez for the 2nd Annual CSEDM Data Challenge - Track 2 <br> loosely based on Yang Azure's naive_model

</p>

<p>

<b> Install/Importing ML Libraries </b>

</p>

```{r}
library(tidyverse)
library(BBmisc)
library(caret)
```

<p>

Loading Student Training Data

</P>

```{r}
# Main path to data set
# base_dir <- "~/spring2022/cs496/cs496_project" 
base_dir <- "~/csedm/cs496-csedm-main"
# path to subject.csv from the data set dir
subject_path  <- "Subject.csv"

# combine paths
train_subjectID_and_grades_path <- paste(base_dir, subject_path, sep="/")

# read data from subject.csv
train_subjectID_and_grades_data <- read_csv(train_subjectID_and_grades_path)
```

<p>

View Table

</p>

```{r}

train_subjectID_and_grades_data
names(train_subjectID_and_grades_data)[names(train_subjectID_and_grades_data) == "X-Grade"] <- "x_grade"

# here we separate the labels
training_labels <- train_subjectID_and_grades_data["x_grade"]
student_id  <-  train_subjectID_and_grades_data["SubjectID"]
# training_labels
# student_id
```

<p>

Hand Crafted Feature Extraction Algorithm

</p>

```{r}
find_avg_attempts = function(student_instance, early_dataset){
  
  # This is get the all data related to the current student
  first_half_data <- filter(early_dataset, SubjectID == student_instance)
  
  # find avg number of attempts
  avg_attempts <- median(first_half_data$Attempts)
}
```

```{r}
find_percent_CorrectEventually = function(student_instance, early_dataset){
  
  # This is get the all data related to the current student
  first_half_data <- filter(early_dataset, SubjectID == student_instance) # All of this student's problem entries from the the early part of the semester
  
  # find percent of time student gets the right answer eventually
  PerCorrectEventually <- mean(first_half_data$CorrectEventually) # Average number correct for all of this student's problems that are attempted

  
  # final_grade <- 100
  }
```

```{r}
# get_final_grade = function(student, dataset) {
#   grade <- dataset[student, "X-Grade"]
# }
```

<p>

Load in the Early semester Performance

</p>

```{r}
# loading early data set
early_dataset_path <- paste(base_dir, "Train/early.csv", sep="/")
early_dataset <- read_csv(early_dataset_path)
```

<p>

Feature Extraction on the training data

</p>

```{r}
# find the avg number of attempts 
train_subjectID_and_grades_data <- train_subjectID_and_grades_data %>% 
  rowwise()  %>%
  mutate(avg_attempts = find_avg_attempts(SubjectID, early_dataset))

# find the percent of time student gets the right answer, eventually
train_subjectID_and_grades_data <- train_subjectID_and_grades_data %>% 
  rowwise()  %>%
  mutate(per_CorrectEventually = 
           find_percent_CorrectEventually(SubjectID,early_dataset))

# student_id <- student_id %>%
#   rowwise() %>%
#   mutate(final_grade =
#            get_final_grade(SubjectID, train_subjectID_and_grades_data))


student_features <- tibble(train_subjectID_and_grades_data)
# student_features <- select(student_features, -c(SubjectID))
student_features

```
```{r}
model1 <- lm(avg_attempts~x_grade, data=student_features)
model2 <- lm(per_CorrectEventually~x_grade, data = student_features)
attempts_grade_plot <- student_features %>% 
  ggplot(aes(x = avg_attempts, y = x_grade)) +
  geom_point() + geom_smooth(method="lm")
  labs(x = "Average attempts", y = "Final grade")
model3 <- lm(x_grade~avg_attempts+per_CorrectEventually, data=student_features)

  
percent_grade_plot <- student_features %>% 
  ggplot(aes(x = per_CorrectEventually, y = x_grade)) +
  geom_point() + geom_smooth(method="lm")
  labs(x = "Percent correct eventually", y = "Final grade")
  
attempts_grade_plot
percent_grade_plot
s1 <- summary(model1)
s2 <- summary(model2)
s3 <- summary(model3)
s1$r.squared # Correlation between average attempts and final grade
s2$r.squared # Correlation between percent eventually correct and final grade
s3$r.squared 

# look at outliers 
```


```{r}

```


```{r}

```

<p>

Standardize Features

</p>

```{r}
student_features <- 
  student_features %>%
  mutate(across(c("avg_attempts", "per_CorrectEventually"), normalize))

student_features
```

<p>

Train Simple ML Model

</p>

```{r}

```

<p>

Evaluation - Calculate Metrics

</p>

```{r}

```