Skip to content
This repository was archived by the owner on Sep 22, 2025. It is now read-only.

sprint2:NLP baselines (BoW/TF-IDF + LogReg/SVM) #358

Open
Tanvir-ctrl1 wants to merge 1 commit into
base: master
Choose a base branch
from
Open

sprint2:NLP baselines (BoW/TF-IDF + LogReg/SVM) #358

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
384 changes: 384 additions & 0 deletions AI Guardian/Emotional_baseline/emotional_baseline.py
View file
Open in desktop

Large diffs are not rendered by default.

Binary file added AI Guardian/Emotional_baseline/out_final/confusion_matrix.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
85 changes: 85 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/feature_preview.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,85 @@
Sample nonzero features for first test doc

--- BoW (1-gram) ---
patient: 3.0000
oxygen: 2.0000
20mg: 1.0000
2l: 1.0000
85: 1.0000
92: 1.0000
administered: 1.0000
alert: 1.0000
bpm: 1.0000
breathing: 1.0000
cannula: 1.0000
currently: 1.0000
drink: 1.0000
encouraged: 1.0000
furosemide: 1.0000
heart: 1.0000
nasal: 1.0000
rate: 1.0000
remains: 1.0000
responsive: 1.0000
reveal: 1.0000
saturation: 1.0000
signs: 1.0000
supplemental: 1.0000
tolerated: 1.0000
Vocab size: 242

--- TF-IDF (1–2) ---
currently breathing: 0.2249
responsive encouraged: 0.2249
2l nasal: 0.2119
85 bpm: 0.2119
administered patient: 0.2119
rate 85: 0.2119
breathing: 0.2019
saturation 92: 0.2019
85: 0.1936
92: 0.1807
alert responsive: 0.1807
responsive: 0.1807
furosemide 20mg: 0.1706
oxygen 2l: 0.1623
20mg administered: 0.1554
cannula: 0.1554
drink water: 0.1554
encouraged drink: 0.1554
nasal: 0.1554
nasal cannula: 0.1554
supplemental: 0.1554
supplemental oxygen: 0.1554
2l: 0.1523
drink: 0.1466
encouraged: 0.1371
Vocab size: 745

--- TF-IDF (1–3) ---
alert responsive encouraged: 0.1876
currently breathing: 0.1876
responsive encouraged: 0.1876
2l nasal: 0.1767
2l nasal cannula: 0.1767
85 bpm: 0.1767
administered patient: 0.1767
heart rate 85: 0.1767
oxygen 2l nasal: 0.1767
patient alert responsive: 0.1767
rate 85: 0.1767
rate 85 bpm: 0.1767
breathing: 0.1683
oxygen saturation 92: 0.1683
saturation 92: 0.1683
85: 0.1615
92: 0.1506
alert responsive: 0.1506
furosemide 20mg administered: 0.1506
responsive: 0.1506
furosemide 20mg: 0.1422
oxygen 2l: 0.1354
supplemental oxygen 2l: 0.1354
20mg administered: 0.1296
cannula: 0.1296
Vocab size: 1331
6 changes: 6 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/lexicon_preview.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,6 @@
pos_count,neg_count,pos_rate,neg_rate,anger_count,joy_count,sadness_count,fear_count,surprise_count,disgust_count,anger_rate,joy_rate,sadness_rate,fear_rate,surprise_rate,disgust_rate
0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
1.0,0.0,0.02,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
1.0,0.0,0.02564102564102564,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
1.0,1.0,0.025,0.025,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
1.0,1.0,0.025,0.025,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
23 changes: 23 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/settings.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
{
"csv": "New AI spreadsheet - Sheet1.csv",
"text_col": "nursingNote",
"label_col": "state",
"group_col": null,
"test_size": 0.2,
"min_df": 2,
"n_splits": 5,
"with_lexicon": true,
"extra_label_words": "",
"stripped_tokens_effective": [
"comfort",
"comfortable",
"normal",
"normality",
"normally",
"sick",
"sickness",
"uncomfort",
"uncomfortable"
],
"seed": 42
}
21 changes: 21 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/sprint2_cv_results.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,21 @@
Fold,Features,Model,Accuracy,F1_macro
1,TFIDF(1–2) + lex,LogReg,0.975,0.9463203463203463
1,TFIDF(1–2) + lex,LinearSVM,0.975,0.9463203463203463
2,TFIDF(1–2) + lex,LogReg,0.975,0.9274853801169591
2,TFIDF(1–2) + lex,LinearSVM,0.975,0.9274853801169591
3,TFIDF(1–2) + lex,LogReg,0.975,0.9270440251572327
3,TFIDF(1–2) + lex,LinearSVM,0.975,0.9270440251572327
4,TFIDF(1–2) + lex,LogReg,0.95,0.9538461538461539
4,TFIDF(1–2) + lex,LinearSVM,0.975,0.9775910364145659
5,TFIDF(1–2) + lex,LogReg,1.0,1.0
5,TFIDF(1–2) + lex,LinearSVM,1.0,1.0
1,TFIDF(1–3) + lex,LogReg,0.975,0.9463203463203463
1,TFIDF(1–3) + lex,LinearSVM,0.975,0.9463203463203463
2,TFIDF(1–3) + lex,LogReg,0.975,0.9274853801169591
2,TFIDF(1–3) + lex,LinearSVM,0.975,0.9274853801169591
3,TFIDF(1–3) + lex,LogReg,0.975,0.9270440251572327
3,TFIDF(1–3) + lex,LinearSVM,0.975,0.9270440251572327
4,TFIDF(1–3) + lex,LogReg,0.95,0.9538461538461539
4,TFIDF(1–3) + lex,LinearSVM,0.975,0.9775910364145659
5,TFIDF(1–3) + lex,LogReg,1.0,1.0
5,TFIDF(1–3) + lex,LinearSVM,1.0,1.0
5 changes: 5 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/sprint2_cv_summary.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
Features,Model,Accuracy_mean,Accuracy_std,F1_mean,F1_std
TFIDF(1–2) + lex,LinearSVM,0.9800000000000001,0.011180339887498959,0.9556881576018208,0.03219826993246655
TFIDF(1–3) + lex,LinearSVM,0.9800000000000001,0.011180339887498959,0.9556881576018208,0.03219826993246655
TFIDF(1–2) + lex,LogReg,0.975,0.017677669529663705,0.9509391810881385,0.029823679075897343
TFIDF(1–3) + lex,LogReg,0.975,0.017677669529663705,0.9509391810881385,0.029823679075897343
64 changes: 64 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/sprint2_report.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
# Sprint 2: Basic NLP Features & Baselines (Leakage-Safe)

- **CSV:** New AI spreadsheet - Sheet1.csv

- **Text column:** `nursingNote` | **Label column:** `state`

- **Groups:** groups by normalized text

- **Dedup:** exact duplicates removed after scrubbing

- **Leakage guard:** stripped tokens ['comfort', 'comfortable', 'normal', 'normality', 'normally', 'sick', 'sickness', 'uncomfort', 'uncomfortable']


## Feature extraction

- BoW (1-gram), min_df=2

- TF-IDF with bigrams & trigrams (captures phrases like *not happy*, *very tired*).

- Lexicon counts: pos/neg + six emotions (anger, joy, sadness, fear, surprise, disgust).


## Baseline models
- Logistic Regression (balanced)
- Linear SVM (balanced)


## Single-split results (sorted by Macro-F1)


Features Model Accuracy F1_macro
BoW(1) + lex LogReg 1.000 1.000000
BoW(1) + lex LinearSVM 1.000 1.000000
TFIDF(1–2) + lex LinearSVM 0.950 0.906020
TFIDF(1–3) + lex LinearSVM 0.950 0.906020
TFIDF(1–3) + lex LogReg 0.925 0.884162
TFIDF(1–2) + lex LogReg 0.900 0.822402



**Best (single split):** BoW(1) + lex + LogReg

Accuracy: **1.000** | Macro-F1: **1.000**


**Confusion matrix** saved to `confusion_matrix.png`.


## 5-fold CV (GroupKFold; mean ± std)


Features Model Accuracy F1_macro
TFIDF(1–2) + lex LinearSVM 0.98 ± 0.011 0.956 ± 0.032
TFIDF(1–3) + lex LinearSVM 0.98 ± 0.011 0.956 ± 0.032
TFIDF(1–2) + lex LogReg 0.975 ± 0.018 0.951 ± 0.03
TFIDF(1–3) + lex LogReg 0.975 ± 0.018 0.951 ± 0.03



## Environment

- python: `/opt/anaconda3/envs/sprint2nlp/bin/python`

- numpy: 2.3.3 | pandas: 2.3.2 | sklearn: 1.7.2
7 changes: 7 additions & 0 deletions AI Guardian/Emotional_baseline/out_final/sprint2_results.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
Features,Model,Accuracy,F1_macro
BoW(1) + lex,LogReg,1.0,1.0
BoW(1) + lex,LinearSVM,1.0,1.0
TFIDF(1–2) + lex,LinearSVM,0.95,0.906020066889632
TFIDF(1–3) + lex,LinearSVM,0.95,0.906020066889632
TFIDF(1–3) + lex,LogReg,0.925,0.8841623785020012
TFIDF(1–2) + lex,LogReg,0.9,0.8224019167415394