Skip to content

Commit 09c80e2

Browse files
committed
model_card update'
1 parent d13f60b commit 09c80e2

File tree

1 file changed

+10
-3
lines changed

1 file changed

+10
-3
lines changed

model_card.md

Lines changed: 10 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,26 @@
77
- Model type: python "RandomForestClassifier" (default parameters)
88

99
## Intended Use
10-
- Classifies people who make more or less than 50k per year into different buckets
10+
- This model is used to predict the income level of individuals based off attiributes such as demographic and employment-related featues. These values are then used to predict an individuals salary being above or below 50k.
1111

1212
## Training Data
13-
- US cencus data provided by Udacity
13+
- US cencus data provided by Udacity, its source is located here: https: (// archive.ics.uci.edu/ml/datasets/census+income).
14+
- This data contains information pertaining to individuals, it was from 1994 Census Bureau data from the UCI Machine Learning Repository.
1415

1516
## Evaluation Data
16-
- US cencus data provided by Udacity
17+
- The data was split to test and train the model at 20% and 80%.
1718

1819
## Metrics
20+
- A random Forest Classifier model was used with the default parameters.
21+
- This model achieved the metric scores below:
1922
- Metric Performance: Precision: 0.7369 | Recall: 0.6294 | F1: 0.6789
2023

2124
## Ethical Considerations
2225
- This model contains features and variables that should not be used for evaluation.
26+
- This model would not be considered a fair representation because it uses demgraphics that include sage, sex and race, and therefore should nmot be used in a descision making process when evaluating individuals.
2327

2428
## Caveats and Recommendations
2529
- Default parameters were used, for a more accurate result other models and parameters could be used.
30+
- The Census database used is from 1994, the data provided is most likely out of date and inacurate.
31+
- Individuals that are under represented in these demographics may not be acuratly portrayed.
32+

0 commit comments

Comments
 (0)