stau.space
this repo has no description
feat: New plots.
Signed-off-by: Sona Tau Estrada Rivera <sona@stau.space>
+138
-18
+43
code/synonym_plots.R
+43
code/synonym_plots.R
···
1
+
library(ggplot2)
2
+
library(reshape2)
3
+
library(viridis)
4
+
5
+
data <- read.csv("data/synonym_thresholds.csv")
6
+
mat <- data.frame(
7
+
data $ min,
8
+
data $ q1,
9
+
data $ median,
10
+
data $ mean,
11
+
data $ q3,
12
+
data $ max
13
+
) |> as.matrix()
14
+
15
+
rownames(mat) <- list(
16
+
"Google News",
17
+
"GloVe 300-D",
18
+
"GloVe 200-D",
19
+
"GloVe 100-D",
20
+
"GloVe 50-D"
21
+
) |> as.character()
22
+
23
+
colnames(mat) <- list(
24
+
"Minimum",
25
+
"First Quartile",
26
+
"Median",
27
+
"Mean",
28
+
"Third Quartile",
29
+
"Max"
30
+
)
31
+
32
+
ld <- melt(mat)
33
+
ld <- ld[ld $ value != 0,]
34
+
35
+
pdf("synonyms.pdf")
36
+
ggplot(ld, aes(x = Var2, y = Var1)) +
37
+
geom_raster(aes(fill = value)) +
38
+
scale_fill_viridis_c() +
39
+
geom_text(aes(label = value)) +
40
+
labs(x = "Quartiles", y = "Models") +
41
+
theme_light() +
42
+
theme(aspect.ratio = 1)
43
+
dev.off()
+93
-18
code/variable_analysis.R
+93
-18
code/variable_analysis.R
···
12
12
library(SnowballC) # wordStem
13
13
# install.packages("gglot2")
14
14
library(ggplot2) # ggplot
15
+
library(viridis) # scale_viridis_d
16
+
library(tidyverse) # ????
15
17
16
18
17
19
load("data/stem_AMS.Rda")
···
27
29
pdf("plots_stem.pdf", onefile = TRUE)
28
30
for (data in data_sources) {
29
31
data <- data[complete.cases(data),]
30
-
print(ggplot(data, aes(x = reorder(word, n), y = n)) +
32
+
print(
33
+
ggplot(data, aes(x = reorder(word, n), y = n)) +
31
34
geom_col() +
32
35
coord_flip() +
33
-
labs(title = paste("Most frequent words in", names[[i]], "surveys")))
36
+
labs(
37
+
title = paste("Most frequent words in", names[[i]], "surveys"),
38
+
x = "Occurrences",
39
+
y = "Words"
40
+
) +
41
+
theme_minimal() +
42
+
theme(aspect.ratio = 1)
43
+
)
34
44
i <- i + 1
35
45
}
36
46
dev.off()
···
66
76
load("data/google_sema_IPEDS.Rda")
67
77
sema_count_google_ipeds <- sema_count
68
78
79
+
get_df <- function(str) {
80
+
data.frame(
81
+
model = list("GloVe 50D", "GloVe 100D", "GloVe 200D", "GloVe 300D", "Google News", "GloVe 50D", "GloVe 100D", "GloVe 200D", "GloVe 300D", "Google News", "GloVe 50D", "GloVe 100D", "GloVe 200D", "GloVe 300D", "Google News") |> as.character(),
82
+
title = list("IPEDS", "IPEDS", "IPEDS", "IPEDS", "IPEDS", "CBMS", "CBMS", "CBMS", "CBMS", "CBMS", "AMS", "AMS", "AMS", "AMS", "AMS") |> as.character(),
83
+
n = list(
84
+
sema_count_glove_50d_ipeds[sema_count_glove_50d_ipeds $ word_category == str,] $ n |> sum(na.rm = TRUE),
85
+
sema_count_glove_300d_ipeds[sema_count_glove_300d_ipeds $ word_category == str,] $ n |> sum(na.rm = TRUE),
86
+
sema_count_glove_200d_ipeds[sema_count_glove_200d_ipeds $ word_category == str,] $ n |> sum(na.rm = TRUE),
87
+
sema_count_glove_100d_ipeds[sema_count_glove_100d_ipeds $ word_category == str,] $ n |> sum(na.rm = TRUE),
88
+
sema_count_google_ipeds[sema_count_google_ipeds $ word_category == str,] $ n |> sum(na.rm = TRUE),
89
+
sema_count_glove_50d_cbms[sema_count_glove_50d_cbms $ word_category == str,] $ n |> sum(na.rm = TRUE),
90
+
sema_count_glove_300d_cbms[sema_count_glove_300d_cbms $ word_category == str,] $ n |> sum(na.rm = TRUE),
91
+
sema_count_glove_200d_cbms[sema_count_glove_200d_cbms $ word_category == str,] $ n |> sum(na.rm = TRUE),
92
+
sema_count_glove_100d_cbms[sema_count_glove_100d_cbms $ word_category == str,] $ n |> sum(na.rm = TRUE),
93
+
sema_count_google_cbms[sema_count_google_cbms $ word_category == str,] $ n |> sum(na.rm = TRUE),
94
+
sema_count_glove_50d_ams[sema_count_glove_50d_ams $ word_category == str,] $ n |> sum(na.rm = TRUE),
95
+
sema_count_glove_300d_ams[sema_count_glove_300d_ams $ word_category == str,] $ n |> sum(na.rm = TRUE),
96
+
sema_count_glove_200d_ams[sema_count_glove_200d_ams $ word_category == str,] $ n |> sum(na.rm = TRUE),
97
+
sema_count_glove_100d_ams[sema_count_glove_100d_ams $ word_category == str,] $ n |> sum(na.rm = TRUE),
98
+
sema_count_google_ams[sema_count_google_ams $ word_category == str,] $ n |> sum(na.rm = TRUE)
99
+
) |> as.numeric()
100
+
)
101
+
}
102
+
103
+
# ------------------------------ lgbt ------------------------------
104
+
105
+
lgbt_df <- get_df("lgbt")
106
+
107
+
race_ethn_df <- get_df("race/ethnicity")
108
+
109
+
women_df <- get_df("women")
110
+
111
+
disabilities_df <- get_df("disabilities")
112
+
113
+
i <- 1
114
+
tags <- list("LGBT", "Race/Ethnicity", "Women", "Disabilities") |> as.character()
115
+
pdf("tags.pdf")
116
+
for (df in list(lgbt_df, race_ethn_df, women_df, disabilities_df)) {
117
+
tag_name <- tags[i]
118
+
print(
119
+
ggplot(df, aes(fill = model, x = title, y = n)) +
120
+
geom_bar(position = "dodge", stat = "identity") +
121
+
scale_fill_viridis_d() +
122
+
labs(
123
+
x = "Data Source",
124
+
y = "Occurrences"
125
+
) +
126
+
theme_light() +
127
+
theme(aspect.ratio = 1)
128
+
)
129
+
i <- i + 1
130
+
}
131
+
dev.off()
132
+
133
+
# women
134
+
# race/ethnicity
135
+
# disabilities
136
+
69
137
data_sources <- list(sema_count_glove_100d_ams,sema_count_glove_100d_cbms,sema_count_glove_100d_ipeds,sema_count_glove_200d_ams,sema_count_glove_200d_cbms,sema_count_glove_200d_ipeds,sema_count_glove_300d_ams,sema_count_glove_300d_cbms,sema_count_glove_300d_ipeds,sema_count_glove_50d_ams,sema_count_glove_50d_cbms,sema_count_glove_50d_ipeds,sema_count_google_ams,sema_count_google_cbms,sema_count_google_ipeds)
70
138
names <- list(
71
-
"AMS surveys w/ 100-D Glove model",
72
-
"CBMS surveys w/ 100-D Glove model",
73
-
"IPEDS surveys w/ 100-D Glove model",
74
-
"AMS surveys w/ 200-D Glove model",
75
-
"CBMS surveys w/ 200-D Glove model",
76
-
"IPEDS surveys w/ 200-D Glove model",
77
-
"AMS surveys w/ 300-D Glove model",
78
-
"CBMS surveys w/ 300-D Glove model",
79
-
"IPEDS surveys w/ 300-D Glove model",
80
-
"AMS surveys w/ 50-D Glove model",
81
-
"CBMS surveys w/ 50-D Glove model",
82
-
"IPEDS surveys w/ 50-D Glove model",
139
+
"AMS surveys w/ 100-D GloVe model",
140
+
"CBMS surveys w/ 100-D GloVe model",
141
+
"IPEDS surveys w/ 100-D GloVe model",
142
+
"AMS surveys w/ 200-D GloVe model",
143
+
"CBMS surveys w/ 200-D GloVe model",
144
+
"IPEDS surveys w/ 200-D GloVe model",
145
+
"AMS surveys w/ 300-D GloVe model",
146
+
"CBMS surveys w/ 300-D GloVe model",
147
+
"IPEDS surveys w/ 300-D GloVe model",
148
+
"AMS surveys w/ 50-D GloVe model",
149
+
"CBMS surveys w/ 50-D GloVe model",
150
+
"IPEDS surveys w/ 50-D GloVe model",
83
151
"AMS surveys w/ Google News model",
84
152
"CBMS surveys w/ Google News model",
85
153
"IPEDS surveys w/ Google News model"
···
89
157
pdf("plots_sema.pdf", onefile = TRUE)
90
158
for (data in data_sources) {
91
159
data <- data[complete.cases(data),]
92
-
print(ggplot(data, aes(x = reorder(tag, n), y = n)) +
93
-
geom_col() +
94
-
coord_flip() +
95
-
labs(title = paste("Most frequent tags in", names[[i]])))
160
+
print(
161
+
ggplot(data) +
162
+
geom_bar(aes(y = word_category, fill = word_category)) +
163
+
labs(
164
+
title = paste("Most frequent tags in", names[[i]]),
165
+
x = "Occurrences",
166
+
y = "Tags"
167
+
) +
168
+
theme_minimal() +
169
+
theme(aspect.ratio = 1)
170
+
)
96
171
i <- i + 1
97
172
}
98
173
dev.off()