|
|
@@ -1,133 +0,0 @@
|
|
|
----
|
|
|
-title: "Solutions for Guided Project: Exploring NYC Schools Survey Data"
|
|
|
-author: "Rose Martin"
|
|
|
-data: "January 22, 2019"
|
|
|
-output: html_document
|
|
|
----
|
|
|
-
|
|
|
-**Here are suggested solutions to the questions in the Data Cleaning With R Guided Project: Exploring NYC Schools Survey Data.**
|
|
|
-
|
|
|
-Load the packages you'll need for your analysis
|
|
|
-
|
|
|
-```{r}
|
|
|
-library(readr)
|
|
|
-library(dplyr)
|
|
|
-library(stringr)
|
|
|
-library(purrr)
|
|
|
-library(tidyr)
|
|
|
-library(ggplot2)
|
|
|
-```
|
|
|
-
|
|
|
-Import the data into R.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined <- read_csv("combined.csv")
|
|
|
-survey <- read_tsv("survey_all.txt")
|
|
|
-survey_d75 <- read_tsv("survey_d75.txt")
|
|
|
-```
|
|
|
-
|
|
|
-Filter `survey` data to include only high schools and select columns needed for analysis based on the data dictionary.
|
|
|
-
|
|
|
-```{r}
|
|
|
-survey_select <- survey %>%
|
|
|
- filter(schooltype == "High School") %>%
|
|
|
- select(dbn:aca_tot_11)
|
|
|
-```
|
|
|
-
|
|
|
-Select columns needed for analysis from `survey_d75`.
|
|
|
-
|
|
|
-```{r}
|
|
|
-survey_d75_select <- survey_d75 %>%
|
|
|
- select(dbn:aca_tot_11)
|
|
|
-```
|
|
|
-
|
|
|
-Combine `survey` and `survey_d75` data frames.
|
|
|
-
|
|
|
-```{r}
|
|
|
-survey_total <- survey_select %>%
|
|
|
- bind_rows(survey_d75_select)
|
|
|
-```
|
|
|
-
|
|
|
-Rename `survey_total` variable `dbn` to `DBN` so can use as key to join with the `combined` data frame.
|
|
|
-
|
|
|
-```{r}
|
|
|
-survey_total <- survey_total %>%
|
|
|
- rename(DBN = dbn)
|
|
|
-```
|
|
|
-
|
|
|
-Join the `combined` and `survey_total` data frames. Use `left_join()` to keep only survey data that correspond to schools for which we have data in `combined`.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined_survey <- combined %>%
|
|
|
- left_join(survey_total, by = "DBN")
|
|
|
-```
|
|
|
-
|
|
|
-Create a correlation matrix to look for interesting relationships between pairs of variables in `combined_survey` and convert it to a tibble so it's easier to work with using tidyverse tools.
|
|
|
-
|
|
|
-```{r}
|
|
|
-cor_mat <- combined_survey %>% ## interesting relationshipsS
|
|
|
- select(avg_sat_score, saf_p_11:aca_tot_11) %>%
|
|
|
- cor(use = "pairwise.complete.obs")
|
|
|
-
|
|
|
-cor_tib <- cor_mat %>%
|
|
|
- as_tibble(rownames = "variable")
|
|
|
-```
|
|
|
-
|
|
|
-Look for correlations of other variables with `avg_sat_score` that are greater than 0.25 or less than -0.25 (strong correlations).
|
|
|
-
|
|
|
-```{r}
|
|
|
-strong_cors <- cor_tib %>%
|
|
|
- select(variable, avg_sat_score) %>%
|
|
|
- filter(avg_sat_score > 0.25 | avg_sat_score < -0.25)
|
|
|
-```
|
|
|
-
|
|
|
-Make scatter plots of those variables with `avg_sat_score` to examine relationships more closely.
|
|
|
-
|
|
|
-```{r}
|
|
|
-create_scatter <- function(x, y) {
|
|
|
- ggplot(data = combined_survey) +
|
|
|
- aes_string(x = x, y = y) +
|
|
|
- geom_point(alpha = 0.3) +
|
|
|
- theme(panel.background = element_rect(fill = "white"))
|
|
|
-}
|
|
|
-
|
|
|
-x_var <- strong_cors$variable[2:5]
|
|
|
-y_var <- "avg_sat_score"
|
|
|
-
|
|
|
-map2(x_var, y_var, create_scatter)
|
|
|
-```
|
|
|
-
|
|
|
-Reshape the data so that you can investigate differences in student, parent, and teacher responses to survey questions.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined_survey_gather <- combined_survey %>%
|
|
|
- gather(key = "survey_question", value = score, saf_p_11:aca_tot_11)
|
|
|
-```
|
|
|
-
|
|
|
-Use `str_sub()` to create new variables, `response_type` and `question`, from the `survey_question` variable.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined_survey_gather <- combined_survey_gather %>%
|
|
|
- mutate(response_type = str_sub(survey_question, 4, 6)) %>%
|
|
|
- mutate(question = str_sub(survey_question, 1, 3))
|
|
|
-```
|
|
|
-
|
|
|
-Replace `response_type` variable values with names "parent", "teacher", "student", "total" using `if_else()` function.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined_survey_gather <- combined_survey_gather %>%
|
|
|
- mutate(response_type = ifelse(response_type == "_p_", "parent",
|
|
|
- ifelse(response_type == "_t_", "teacher",
|
|
|
- ifelse(response_type == "_s_", "student",
|
|
|
- ifelse(response_type == "_to", "total", "NA")))))
|
|
|
-```
|
|
|
-
|
|
|
-Make a boxplot to see if there appear to be differences in how the three groups of responders (parents, students, and teachers) answered the four questions.
|
|
|
-
|
|
|
-```{r}
|
|
|
-combined_survey_gather %>%
|
|
|
- filter(response_type != "total") %>%
|
|
|
- ggplot() +
|
|
|
- aes(x = question, y = score, fill = response_type) +
|
|
|
- geom_boxplot()
|
|
|
-```
|
Rose Martin