Skip to content

Commit 545af38

Browse files
AnFreThAnFreTh
committed
adjust readme with results
1 parent 08a8797 commit 545af38

1 file changed

Lines changed: 42 additions & 40 deletions

File tree

‎README.md‎

Lines changed: 42 additions & 40 deletions
Original file line numberDiff line numberDiff line change
@@ -26,8 +26,6 @@ You can find our paper `Mambular: A Sequential Model for Tabular Deep Learning`
2626
- [Features](#features)
2727
- [Models](#models)
2828
- [Results](#results)
29-
- [Regression Results](#regression-results)
30-
- [Classification Results](#classification-results)
3129
- [Documentation](#documentation)
3230
- [Installation](#installation)
3331
- [Usage](#usage)
@@ -83,49 +81,53 @@ The average rank table over all models and all datasets is given here:
8381

8482
<div align="center">
8583

86-
| **Model** | **Avg. Rank** |
87-
| -------------------- | --------------------------- |
88-
| **Mambular** | **2.083** <sub>±1.037</sub> |
89-
| FT-Transformer | 2.417 <sub>±1.256</sub> |
90-
| XGBoost | 3.167 <sub>±2.577</sub> |
91-
| MambaTab* | 4.333 <sub>±1.374</sub> |
92-
| ResNet | 4.750 <sub>±1.639</sub> |
93-
| TabTransformer | 6.222 <sub>±1.618</sub> |
94-
| MLP | 6.500 <sub>±1.500</sub> |
95-
| MambaTab | 6.583 <sub>±1.801</sub> |
96-
| MambaTab<sup>T</sup> | 7.917 <sub>±1.187</sub> |
84+
<table>
85+
<tr>
86+
<th style="text-align:center;">Model</th>
87+
<th style="text-align:center;">Avg. Rank</th>
88+
</tr>
89+
<tr>
90+
<td style="text-align:center;"><strong style="color:#2e8b57;">Mambular</strong><br><br></td>
91+
<td style="text-align:center;"><strong style="color:#2e8b57;">2.083</strong> <sub>±1.037</sub></td>
92+
</tr>
93+
<tr>
94+
<td style="text-align:center;">FT-Transformer</td>
95+
<td style="text-align:center;">2.417 <sub>±1.256</sub></td>
96+
</tr>
97+
<tr>
98+
<td style="text-align:center;">XGBoost</td>
99+
<td style="text-align:center;">3.167 <sub>±2.577</sub></td>
100+
</tr>
101+
<tr>
102+
<td style="text-align:center;">MambaTab*</td>
103+
<td style="text-align:center;">4.333 <sub>±1.374</sub></td>
104+
</tr>
105+
<tr>
106+
<td style="text-align:center;">ResNet</td>
107+
<td style="text-align:center;">4.750 <sub>±1.639</sub></td>
108+
</tr>
109+
<tr>
110+
<td style="text-align:center;">TabTransformer</td>
111+
<td style="text-align:center;">6.222 <sub>±1.618</sub></td>
112+
</tr>
113+
<tr>
114+
<td style="text-align:center;">MLP</td>
115+
<td style="text-align:center;">6.500 <sub>±1.500</sub></td>
116+
</tr>
117+
<tr>
118+
<td style="text-align:center;">MambaTab</td>
119+
<td style="text-align:center;">6.583 <sub>±1.801</sub></td>
120+
</tr>
121+
<tr>
122+
<td style="text-align:center;">MambaTab<sup>T</sup></td>
123+
<td style="text-align:center;">7.917 <sub>±1.187</sub></td>
124+
</tr>
125+
</table>
97126

98127
</div>
99128

100129

101-
The detailed results for different datasets are given below.
102130

103-
#### Regression Results
104-
The given values are MSE values. All results are achieved with 5-fold CV and the standard deviation is reported.
105-
106-
107-
| Models | DI ↓ | AB ↓ | CA ↓ | WI ↓ | PA ↓ | HS ↓ | CP ↓ |
108-
| -------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- |
109-
| XGBoost | 0.019 <sub>±0.000</sub> | 0.506 <sub>±0.044</sub> | 0.171 <sub>±0.007</sub> | **0.528** <sub>±0.008</sub> | 0.036 <sub>±0.004</sub> | 0.119 <sub>±0.024</sub> | **0.024** <sub>±0.004</sub> |
110-
| FT-Transformer | 0.018 <sub>±0.001</sub> | 0.458 <sub>±0.055</sub> | 0.169 <sub>±0.006</sub> | 0.615 <sub>±0.012</sub> | **0.024** <sub>±0.005</sub> | **0.111** <sub>±0.014</sub> | 0.024 <sub>±0.001</sub> |
111-
| MLP | 0.066 <sub>±0.003</sub> | 0.462 <sub>±0.051</sub> | 0.198 <sub>±0.011</sub> | 0.654 <sub>±0.013</sub> | 0.764 <sub>±0.023</sub> | 0.147 <sub>±0.017</sub> | 0.031 <sub>±0.001</sub> |
112-
| TabTransformer | 0.065 <sub>±0.002</sub> | 0.472 <sub>±0.057</sub> | 0.247 <sub>±0.013</sub> | - | 0.135 <sub>±0.001</sub> | 0.160 <sub>±0.028</sub> | - |
113-
| ResNet | 0.039 <sub>±0.018</sub> | 0.455 <sub>±0.045</sub> | 0.178 <sub>±0.006</sub> | 0.639 <sub>±0.013</sub> | 0.606 <sub>±0.031</sub> | 0.141 <sub>±0.017</sub> | 0.030 <sub>±0.002</sub> |
114-
| MambaTab* | 0.040 <sub>±0.008</sub> | 0.455 <sub>±0.043</sub> | 0.180 <sub>±0.008</sub> | 0.601 <sub>±0.010</sub> | 0.571 <sub>±0.021</sub> | 0.122 <sub>±0.017</sub> | 0.030 <sub>±0.002</sub> |
115-
| **Mambular** | **0.018** <sub>±0.000</sub> | **0.452** <sub>±0.043</sub> | **0.167** <sub>±0.011</sub> | 0.628 <sub>±0.010</sub> | 0.035 <sub>±0.005</sub> | 0.132 <sub>±0.020</sub> | 0.027 <sub>±0.002</sub> |
116-
117-
#### Classification Results
118-
The table reports AUC values for binary classification tasks
119-
120-
| Models | BA ↑ | AD ↑ | CH ↑ | FI ↑ | MA ↑ |
121-
| -------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- | --------------------------- |
122-
| XGBoost | **0.928** <sub>±0.004</sub> | **0.929** <sub>±0.002</sub> | 0.845 <sub>±0.008</sub> | 0.774 <sub>±0.009</sub> | **0.922** <sub>±0.002</sub> |
123-
| FT-Transformer | 0.926 <sub>±0.003</sub> | 0.926 <sub>±0.002</sub> | **0.863** <sub>±0.007</sub> | 0.792 <sub>±0.011</sub> | 0.916 <sub>±0.003</sub> |
124-
| MLP | 0.895 <sub>±0.004</sub> | 0.914 <sub>±0.002</sub> | 0.840 <sub>±0.005</sub> | 0.793 <sub>±0.011</sub> | 0.886 <sub>±0.003</sub> |
125-
| TabTransformer | 0.921 <sub>±0.004</sub> | 0.912 <sub>±0.002</sub> | 0.835 <sub>±0.007</sub> | - | 0.910 <sub>±0.002</sub> |
126-
| ResNet | 0.896 <sub>±0.006</sub> | 0.917 <sub>±0.002</sub> | 0.841 <sub>±0.006</sub> | 0.793 <sub>±0.013</sub> | 0.889 <sub>±0.003</sub> |
127-
| MambaTab* | 0.900 <sub>±0.004</sub> | 0.916 <sub>±0.003</sub> | 0.846 <sub>±0.007</sub> | 0.792 <sub>±0.011</sub> | 0.890 <sub>±0.003</sub> |
128-
| **Mambular** | 0.927 <sub>±0.006</sub> | 0.928 <sub>±0.002</sub> | 0.856 <sub>±0.004</sub> | **0.795** <sub>±0.011</sub> | 0.917 <sub>±0.003</sub> |
129131

130132

131133
## Documentation

0 commit comments

Comments
 (0)