Fixing Inconsistency in Classification Model -- 2

I'm facing an issue with my classification model where the training features are not consistent with the test features.

- Problem: The inconsistency arises during the feature selection phase.
- Solution: I need an expert who can rectify the feature selection process to ensure consistency across training and testing phases.

Notes:
1.) I have trained and saved models for each league
2.) The accuracy can be better
3.) I made the live prediction part but am getting some errors with features
"Error: X has 161 features, but LogisticRegression is expecting 155 features as input."

the end result is to see predictions in the form:
PREDICTIONS DONE: 2024-09-09

? **Premier League**:
- ⚽ **Southampton** ? **Man United**: Under 2.5 Goals (61.63% chance)
- ⚽ **Brighton** ? **Ipswich**: Over 2.5 Goals! ? (85.26% chance)
- ⚽ **Crystal Palace** ? **Leicester**: Over 2.5 Goals! ? (71.3% chance)
- ⚽ **Fulham** ? **West Ham**: Over 2.5 Goals! ? (75.39% chance)
- ⚽ **Liverpool** ? **Nott'm Forest**: Over 2.5 Goals! ? (56.05% chance)
- ⚽ **Man City** ? **Brentford**: Over 2.5 Goals! ? (86.87% chance)
- ⚽ **Aston Villa** ? **Everton**: Under 2.5 Goals (59.1% chance)
- ⚽ **Bournemouth** ? **Chelsea**: Over 2.5 Goals! ? (89.64% chance)
- ⚽ **Tottenham** ? **Arsenal**: Over 2.5 Goals! ? (84.23% chance)
- ⚽ **Wolves** ? **Newcastle**: Over 2.5 Goals! ? (55.86% chance)

Prefereably in a dataframe