Datasets:
Upload load_dataset.py
Browse files- load_dataset.py +183 -0
load_dataset.py
ADDED
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
"""
|
2 |
+
Simple script to load the Misery Index Dataset using pandas or datasets library.
|
3 |
+
"""
|
4 |
+
|
5 |
+
import pandas as pd
|
6 |
+
from typing import Dict, List, Optional
|
7 |
+
|
8 |
+
def load_misery_dataset(file_path: str = "Misery_Data.csv") -> pd.DataFrame:
|
9 |
+
"""
|
10 |
+
Load the Misery Index Dataset from CSV file.
|
11 |
+
|
12 |
+
Args:
|
13 |
+
file_path: Path to the CSV file (default: "Misery_Data.csv")
|
14 |
+
|
15 |
+
Returns:
|
16 |
+
pandas.DataFrame with cleaned column names and proper data types
|
17 |
+
"""
|
18 |
+
df = pd.read_csv(file_path)
|
19 |
+
|
20 |
+
# Rename columns to be more user-friendly
|
21 |
+
column_mapping = {
|
22 |
+
"Ep #": "episode",
|
23 |
+
"Misery": "scenario",
|
24 |
+
"Score": "misery_score",
|
25 |
+
"VNTO": "vnto",
|
26 |
+
"Reward": "reward",
|
27 |
+
"Win": "win",
|
28 |
+
"Comments": "comments",
|
29 |
+
"question_tag": "question_tag",
|
30 |
+
"level": "level"
|
31 |
+
}
|
32 |
+
|
33 |
+
df = df.rename(columns=column_mapping)
|
34 |
+
|
35 |
+
# Convert misery_score to numeric
|
36 |
+
df["misery_score"] = pd.to_numeric(df["misery_score"], errors="coerce")
|
37 |
+
|
38 |
+
# Convert reward to numeric, handling empty values
|
39 |
+
df["reward"] = pd.to_numeric(df["reward"], errors="coerce").fillna(0).astype(int)
|
40 |
+
|
41 |
+
# Clean up string columns
|
42 |
+
string_columns = ["episode", "scenario", "vnto", "win", "comments", "question_tag", "level"]
|
43 |
+
for col in string_columns:
|
44 |
+
if col in df.columns:
|
45 |
+
df[col] = df[col].astype(str).str.strip()
|
46 |
+
df[col] = df[col].replace("nan", "")
|
47 |
+
|
48 |
+
return df
|
49 |
+
|
50 |
+
def get_dataset_statistics(df: pd.DataFrame) -> Dict:
|
51 |
+
"""
|
52 |
+
Get basic statistics about the dataset.
|
53 |
+
|
54 |
+
Args:
|
55 |
+
df: DataFrame containing the dataset
|
56 |
+
|
57 |
+
Returns:
|
58 |
+
Dictionary with dataset statistics
|
59 |
+
"""
|
60 |
+
stats = {
|
61 |
+
"total_samples": len(df),
|
62 |
+
"mean_misery": df["misery_score"].mean(),
|
63 |
+
"std_misery": df["misery_score"].std(),
|
64 |
+
"min_misery": df["misery_score"].min(),
|
65 |
+
"max_misery": df["misery_score"].max(),
|
66 |
+
"percentiles": {
|
67 |
+
"25th": df["misery_score"].quantile(0.25),
|
68 |
+
"50th": df["misery_score"].quantile(0.50),
|
69 |
+
"75th": df["misery_score"].quantile(0.75),
|
70 |
+
},
|
71 |
+
"vnto_types": df["vnto"].value_counts().to_dict() if "vnto" in df.columns else {},
|
72 |
+
"episodes": df["episode"].nunique(),
|
73 |
+
"question_tags": df["question_tag"].value_counts().to_dict() if "question_tag" in df.columns else {},
|
74 |
+
}
|
75 |
+
return stats
|
76 |
+
|
77 |
+
def filter_by_vnto(df: pd.DataFrame, vnto_type: str) -> pd.DataFrame:
|
78 |
+
"""
|
79 |
+
Filter dataset by VNTO type.
|
80 |
+
|
81 |
+
Args:
|
82 |
+
df: DataFrame containing the dataset
|
83 |
+
vnto_type: VNTO type to filter by (T, V, N, O, P)
|
84 |
+
|
85 |
+
Returns:
|
86 |
+
Filtered DataFrame
|
87 |
+
"""
|
88 |
+
return df[df["vnto"] == vnto_type].copy()
|
89 |
+
|
90 |
+
def filter_by_misery_range(df: pd.DataFrame, min_score: float = 0, max_score: float = 100) -> pd.DataFrame:
|
91 |
+
"""
|
92 |
+
Filter dataset by misery score range.
|
93 |
+
|
94 |
+
Args:
|
95 |
+
df: DataFrame containing the dataset
|
96 |
+
min_score: Minimum misery score (inclusive)
|
97 |
+
max_score: Maximum misery score (inclusive)
|
98 |
+
|
99 |
+
Returns:
|
100 |
+
Filtered DataFrame
|
101 |
+
"""
|
102 |
+
return df[(df["misery_score"] >= min_score) & (df["misery_score"] <= max_score)].copy()
|
103 |
+
|
104 |
+
def get_sample_scenarios(df: pd.DataFrame, vnto_type: Optional[str] = None, n: int = 5) -> List[Dict]:
|
105 |
+
"""
|
106 |
+
Get sample scenarios from the dataset.
|
107 |
+
|
108 |
+
Args:
|
109 |
+
df: DataFrame containing the dataset
|
110 |
+
vnto_type: Optional VNTO type to filter by
|
111 |
+
n: Number of samples to return
|
112 |
+
|
113 |
+
Returns:
|
114 |
+
List of dictionaries with scenario information
|
115 |
+
"""
|
116 |
+
if vnto_type:
|
117 |
+
df_filtered = filter_by_vnto(df, vnto_type)
|
118 |
+
else:
|
119 |
+
df_filtered = df
|
120 |
+
|
121 |
+
samples = df_filtered.sample(n=min(n, len(df_filtered)))
|
122 |
+
|
123 |
+
return [
|
124 |
+
{
|
125 |
+
"scenario": row["scenario"],
|
126 |
+
"misery_score": row["misery_score"],
|
127 |
+
"vnto": row["vnto"],
|
128 |
+
"episode": row["episode"]
|
129 |
+
}
|
130 |
+
for _, row in samples.iterrows()
|
131 |
+
]
|
132 |
+
|
133 |
+
def main():
|
134 |
+
"""Example usage of the dataset loading functions."""
|
135 |
+
|
136 |
+
# Load the dataset
|
137 |
+
print("Loading Misery Index Dataset...")
|
138 |
+
df = load_misery_dataset()
|
139 |
+
|
140 |
+
# Get basic statistics
|
141 |
+
stats = get_dataset_statistics(df)
|
142 |
+
print(f"\nDataset Statistics:")
|
143 |
+
print(f"Total samples: {stats['total_samples']}")
|
144 |
+
print(f"Mean misery score: {stats['mean_misery']:.2f}")
|
145 |
+
print(f"Standard deviation: {stats['std_misery']:.2f}")
|
146 |
+
print(f"Score range: {stats['min_misery']}-{stats['max_misery']}")
|
147 |
+
print(f"Number of episodes: {stats['episodes']}")
|
148 |
+
|
149 |
+
print(f"\nPercentiles:")
|
150 |
+
for p, value in stats['percentiles'].items():
|
151 |
+
print(f" {p}: {value:.2f}")
|
152 |
+
|
153 |
+
print(f"\nVNTO Types distribution:")
|
154 |
+
for vnto_type, count in stats['vnto_types'].items():
|
155 |
+
percentage = (count / stats['total_samples']) * 100
|
156 |
+
print(f" {vnto_type}: {count} ({percentage:.1f}%)")
|
157 |
+
|
158 |
+
print(f"\nTop Question Tags:")
|
159 |
+
for tag, count in list(stats['question_tags'].items())[:5]:
|
160 |
+
percentage = (count / stats['total_samples']) * 100
|
161 |
+
print(f" {tag}: {count} ({percentage:.1f}%)")
|
162 |
+
|
163 |
+
# Show some examples
|
164 |
+
print(f"\nSample low misery scenarios (< 30):")
|
165 |
+
low_misery = filter_by_misery_range(df, 0, 30)
|
166 |
+
samples = get_sample_scenarios(low_misery, n=3)
|
167 |
+
for sample in samples:
|
168 |
+
print(f" Score {sample['misery_score']}: {sample['scenario']}")
|
169 |
+
|
170 |
+
print(f"\nSample high misery scenarios (> 80):")
|
171 |
+
high_misery = filter_by_misery_range(df, 80, 100)
|
172 |
+
samples = get_sample_scenarios(high_misery, n=3)
|
173 |
+
for sample in samples:
|
174 |
+
print(f" Score {sample['misery_score']}: {sample['scenario']}")
|
175 |
+
|
176 |
+
print(f"\nSample Video scenarios (VNTO=V):")
|
177 |
+
video_scenarios = filter_by_vnto(df, "V")
|
178 |
+
samples = get_sample_scenarios(video_scenarios, n=3)
|
179 |
+
for sample in samples:
|
180 |
+
print(f" Score {sample['misery_score']}: {sample['scenario']}")
|
181 |
+
|
182 |
+
if __name__ == "__main__":
|
183 |
+
main()
|