General
PromptBeginner5 minmarkdown
# STEP 12 — MICE / IterativeImputer (most powerful
use when needed)
0
Explore
Find agent skills by outcome
63,729 skills indexed with the new KISS metadata standard.
Showing 24 of 63,729Categories: Cursor-rules, Data, General, Creative, Coding & Debugging
General
PromptBeginner5 minmarkdown
X_train = X_train.drop(columns=drop_cols
errors='ignore')
0
General
PromptBeginner5 minmarkdown
X_train
X_test
0
General
PromptBeginner5 minmarkdown
X
y
0
General
PromptBeginner5 minmarkdown
df.dropna(subset=[TARGET_COL]
axis=0
0
General
PromptBeginner5 minmarkdown
DISGUISED_NULLS = [?
N/A
0
General
PromptBeginner5 minmarkdown
df.replace(DISGUISED_NULLS
np.nan
0
General
PromptBeginner5 minmarkdown
│ Tree-based (XGBoost
LightGBM
0
General
PromptBeginner5 minmarkdown
│ SVM
KNN Classifier: │
0
General
PromptBeginner5 minmarkdown
│ Linear Models (LogReg
LinearReg
0
General
PromptBeginner5 minmarkdown
→ df.dropna(subset=[TARGET_COL]
inplace=True)
0
General
PromptBeginner5 minmarkdown
│ → Use max_iter=10
random_state=42 for reproducibility │
0
General
PromptBeginner5 minmarkdown
│ → Creates: col_was_missing = 1 if NaN
else 0 │
0
General
PromptBeginner5 minmarkdown
For **each column** with missing values
evaluate all three branches simultaneously:
0
General
PromptBeginner5 minmarkdown
**For each flagged column
fill in this analysis card:**
0
General
PromptBeginner5 minmarkdown
│ │ Signs: Missingness correlates with OTHER columns
│
0
General
PromptBeginner5 minmarkdown
missing_report = missing_report.sort_values('Missing_%'
ascending=False)
0
General
PromptBeginner5 minmarkdown
DISGUISED_NULLS = [?
N/A
0
General
PromptBeginner5 minmarkdown
df.replace(DISGUISED_NULLS
np.nan
0
General
PromptBeginner5 minmarkdown
- e.g.
CustomerID must be unique and non-null
0
General
PromptBeginner5 minmarkdown
- e.g.
Age cannot be 0 or negative
0
General
PromptBeginner5 minmarkdown
- e.g.
Price is the target — rows missing it are unusable
0
General
PromptBeginner5 minmarkdown
- Watch for: ?
N/A
0
General
PromptBeginner5 minmarkdown
- Boolean → binary flags (0/1
True/False)
0