Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,6 +1,7 @@
|
|
| 1 |
---
|
| 2 |
language:
|
| 3 |
- en
|
|
|
|
| 4 |
tags:
|
| 5 |
- time-series
|
| 6 |
- classification
|
|
@@ -19,7 +20,7 @@ It classifies test series in a **single forward pass** given labeled training ex
|
|
| 19 |
## Usage
|
| 20 |
|
| 21 |
```bash
|
| 22 |
-
pip install timee
|
| 23 |
```
|
| 24 |
|
| 25 |
```python
|
|
@@ -40,9 +41,21 @@ predictions, probabilities = clf.predict(X_train, y_train, X_test)
|
|
| 40 |
Labels can be any type (`int`, `str`, etc.).
|
| 41 |
Datasets with more than 10 classes are handled automatically via one-vs-rest.
|
| 42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
## Model Details
|
| 44 |
|
| 45 |
-
- **Parameters:**
|
| 46 |
- **Input:** univariate or multivariate time series, any length
|
| 47 |
- **Output:** class probabilities over up to 10 classes (OvR for more)
|
| 48 |
- **Inference:** single forward pass; no dataset-specific adaptation
|
|
|
|
| 1 |
---
|
| 2 |
language:
|
| 3 |
- en
|
| 4 |
+
library_name: timee-ts
|
| 5 |
tags:
|
| 6 |
- time-series
|
| 7 |
- classification
|
|
|
|
| 20 |
## Usage
|
| 21 |
|
| 22 |
```bash
|
| 23 |
+
pip install timee-ts
|
| 24 |
```
|
| 25 |
|
| 26 |
```python
|
|
|
|
| 41 |
Labels can be any type (`int`, `str`, etc.).
|
| 42 |
Datasets with more than 10 classes are handled automatically via one-vs-rest.
|
| 43 |
|
| 44 |
+
## UCR Benchmark Results
|
| 45 |
+
|
| 46 |
+
| Dataset | Classes | Accuracy | ROC AUC |
|
| 47 |
+
|--------------------|---------|----------|---------|
|
| 48 |
+
| ArrowHead | 3 | 76.6 % | 0.964 |
|
| 49 |
+
| ECG5000 | 5 | 95.0 % | 0.952 |
|
| 50 |
+
| GunPoint | 2 | 98.7 % | 0.997 |
|
| 51 |
+
| ItalyPowerDemand | 2 | 96.0 % | 0.993 |
|
| 52 |
+
| TwoPatterns | 4 | 99.8 % | 1.000 |
|
| 53 |
+
|
| 54 |
+
Results use the default 4-member ensemble (interpolate × {256, 512} × {raw, first-difference}).
|
| 55 |
+
|
| 56 |
## Model Details
|
| 57 |
|
| 58 |
+
- **Parameters:** 4,557,322
|
| 59 |
- **Input:** univariate or multivariate time series, any length
|
| 60 |
- **Output:** class probabilities over up to 10 classes (OvR for more)
|
| 61 |
- **Inference:** single forward pass; no dataset-specific adaptation
|