actinet 0.0.dev3__tar.gz → 0.0.dev5__tar.gz

{actinet-0.0.dev3 → actinet-0.0.dev5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: actinet
-Version: 0.0.dev3
+Version: 0.0.dev5
 Summary: Activity detection algorithm compatible with the UK Biobank Accelerometer Dataset
 Home-page: https://github.com/OxWearables/actinet
 Download-URL: https://github.com/OxWearables/actinet
@@ -60,16 +60,16 @@ You are all set! The next time that you want to use `actinet`, open the Anaconda
 
 ```bash
 # Process an AX3 file
-$ actinet sample.cwa
+$ actinet -f sample.cwa
 
 # Or an ActiGraph file
-$ actinet sample.gt3x
+$ actinet -f sample.gt3x
 
 # Or a GENEActiv file
-$ actinet sample.bin
+$ actinet -f sample.bin
 
 # Or a CSV file (see data format below)
-$ actinet sample.csv
+$ actinet -f sample.csv
 ```
 
 ### Troubleshooting
@@ -80,12 +80,34 @@ Some systems may face issues with Java when running the script. If this is your
 conda install -n actinet openjdk=8
 ```
 
+### Offline usage
+
+To use this package offline, one must first download and install the relevant classifier file and model modules.
+This repository offers two ways of doing this.
+
+Run the following code when you have internet access:
+```console
+actinet --cache-classifier
+```
+ 
+Following this, the actinet classifier can be used as standard without internet access, without needing to specify the flags relating to the model repository.
+ 
+Alternatively, you can download or git clone the ssl modules from the [ssl-wearables repository](https://github.com/OxWearables/ssl-wearables).
+
+In addition, you can donwload/prepare a custom classifier file.
+
+Once this is downloaded to an appopriate location, you can run the actinet model using:
+ 
+```console
+actinet -f sample.cwa -c /path/to/classifier.joblib.lzma -m /path/to/ssl-wearables
+```
+
 ### Output files
 
 By default, output files will be stored in a folder named after the input file, `outputs/{filename}/`, created in the current working directory. You can change the output path with the `-o` flag:
 
 ```console
-$ actinet sample.cwa -o /path/to/some/folder/
+$ actinet -f sample.cwa -o /path/to/some/folder/
 
 <Output summary written to: /path/to/some/folder/sample-outputSummary.json>
 <Time series output written to: /path/to/some/folder/sample-timeSeries.csv.gz>
@@ -103,7 +125,7 @@ See [Data Dictionary](https://biobankaccanalysis.readthedocs.io/en/latest/datadi
 To plot the activity profiles, you can use the -p flag:
 
 ```console
-$ actinet sample.cwa -p
+$ actinet -f sample.cwa -p
 <Output plot written to: data/sample-timeSeries-plot.png>
 ```
 
@@ -138,9 +160,9 @@ To process multiple files you can create a text file in Notepad which includes o
 Example text file *commands.txt*:
 
 ```console
-actinet file1.cwa &
-actinet file2.cwa &
-actinet file3.cwa 
+actinet -f file1.cwa &
+actinet -f file2.cwa &
+actinet -f file3.cwa 
 :END
 ````
 
@@ -151,9 +173,9 @@ Once this file is created, run `cmd < commands.txt` from the terminal.
 Create a file *command.sh* with:
 
 ```console
-actinet file1.cwa
-actinet file2.cwa
-actinet file3.cwa
+actinet -f file1.cwa
+actinet -f file2.cwa
+actinet -f file3.cwa
 ```
 
 Then, run `bash command.sh` from the terminal.

{actinet-0.0.dev3 → actinet-0.0.dev5}/README.md

@@ -38,16 +38,16 @@ You are all set! The next time that you want to use `actinet`, open the Anaconda
 
 ```bash
 # Process an AX3 file
-$ actinet sample.cwa
+$ actinet -f sample.cwa
 
 # Or an ActiGraph file
-$ actinet sample.gt3x
+$ actinet -f sample.gt3x
 
 # Or a GENEActiv file
-$ actinet sample.bin
+$ actinet -f sample.bin
 
 # Or a CSV file (see data format below)
-$ actinet sample.csv
+$ actinet -f sample.csv
 ```
 
 ### Troubleshooting
@@ -58,12 +58,34 @@ Some systems may face issues with Java when running the script. If this is your
 conda install -n actinet openjdk=8
 ```
 
+### Offline usage
+
+To use this package offline, one must first download and install the relevant classifier file and model modules.
+This repository offers two ways of doing this.
+
+Run the following code when you have internet access:
+```console
+actinet --cache-classifier
+```
+ 
+Following this, the actinet classifier can be used as standard without internet access, without needing to specify the flags relating to the model repository.
+ 
+Alternatively, you can download or git clone the ssl modules from the [ssl-wearables repository](https://github.com/OxWearables/ssl-wearables).
+
+In addition, you can donwload/prepare a custom classifier file.
+
+Once this is downloaded to an appopriate location, you can run the actinet model using:
+ 
+```console
+actinet -f sample.cwa -c /path/to/classifier.joblib.lzma -m /path/to/ssl-wearables
+```
+
 ### Output files
 
 By default, output files will be stored in a folder named after the input file, `outputs/{filename}/`, created in the current working directory. You can change the output path with the `-o` flag:
 
 ```console
-$ actinet sample.cwa -o /path/to/some/folder/
+$ actinet -f sample.cwa -o /path/to/some/folder/
 
 <Output summary written to: /path/to/some/folder/sample-outputSummary.json>
 <Time series output written to: /path/to/some/folder/sample-timeSeries.csv.gz>
@@ -81,7 +103,7 @@ See [Data Dictionary](https://biobankaccanalysis.readthedocs.io/en/latest/datadi
 To plot the activity profiles, you can use the -p flag:
 
 ```console
-$ actinet sample.cwa -p
+$ actinet -f sample.cwa -p
 <Output plot written to: data/sample-timeSeries-plot.png>
 ```
 
@@ -116,9 +138,9 @@ To process multiple files you can create a text file in Notepad which includes o
 Example text file *commands.txt*:
 
 ```console
-actinet file1.cwa &
-actinet file2.cwa &
-actinet file3.cwa 
+actinet -f file1.cwa &
+actinet -f file2.cwa &
+actinet -f file3.cwa 
 :END
 ````
 
@@ -129,9 +151,9 @@ Once this file is created, run `cmd < commands.txt` from the terminal.
 Create a file *command.sh* with:
 
 ```console
-actinet file1.cwa
-actinet file2.cwa
-actinet file3.cwa
+actinet -f file1.cwa
+actinet -f file2.cwa
+actinet -f file3.cwa
 ```
 
 Then, run `bash command.sh` from the terminal.

{actinet-0.0.dev3 → actinet-0.0.dev5}/setup.py

@@ -55,7 +55,7 @@ def main():
             "scipy==1.10.*",
             "pandas==2.0.*",
             "tqdm==4.64.*",
-            "matplotlib==3.1.*",
+            "matplotlib==3.5.*",
             "joblib==1.2.*",
             "scikit-learn==1.1.1",
             "imbalanced-learn==0.9.1",

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/__init__.py

@@ -4,8 +4,8 @@ __maintainer__ = "Shing Chan"
 __maintainer_email__ = "shing.chan@ndph.ox.ac.uk"
 __license__ = "See LICENSE file."
 
-__model_version__ = "ssl_ukb_c24_rw_20240204"
-__model_md5__ = "c13390ff024e4714eb027aa66160e2eb"
+__classifier_version__ = "ssl_ukb_c24_rw_20240204"
+__classifier_md5__ = "11c3f36348dae37da4f99bd6d810bbb2"
 
 from . import _version
 

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2024-02-05T11:05:44+0000",
+ "date": "2024-02-06T18:32:30+0000",
  "dirty": false,
  "error": null,
- "full-revisionid": "6f3133800a30a331e1a981b9b93937b5fd0e7f1a",
- "version": "0.0.dev3"
+ "full-revisionid": "ae03389ab9965e52140c46a6c43e39799ad1c61b",
+ "version": "0.0.dev5"
 }
 '''  # END VERSION_JSON
 

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/actinet.py

@@ -12,16 +12,15 @@ import joblib
 
 import actipy
 
-from actinet import __model_version__
-from actinet import __model_md5__
+from actinet import __classifier_version__
+from actinet import __classifier_md5__
 from actinet.accPlot import plotTimeSeries
 from actinet.models import ActivityClassifier
 from actinet.sslmodel import SAMPLE_RATE
-from actinet.summarisation import getActivitySummary, ACTIVITY_LABELS
+from actinet.summarisation import getActivitySummary
 from actinet.utils.utils import infer_freq
 
-BASE_URL = "https://zenodo.org/records/10616280/files/"
-
+BASE_URL = "https://zenodo.org/records/10625542/files/"
 
 def main():
 
@@ -29,7 +28,7 @@ def main():
         description="A tool to predict activities from accelerometer data using a self-supervised Resnet 18 model",
         add_help=True,
     )
-    parser.add_argument("filepath", help="Enter file to be processed")
+    parser.add_argument("--filepath", "-f", help="Enter file to be processed")
     parser.add_argument(
         "--outdir",
         "-o",
@@ -37,10 +36,15 @@ def main():
         default="outputs/",
     )
     parser.add_argument(
-        "--model-path", "-m", help="Enter custom model file to use", default=None
+        "--classifier-path",
+        "-c",
+        help="Enter custom acitivty classifier file to use",
+        default=None,
     )
     parser.add_argument(
-        "--force-download", action="store_true", help="Force download of model file"
+        "--force-download",
+        action="store_true",
+        help="Force download of classifier file",
     )
     parser.add_argument(
         "--pytorch-device",
@@ -63,9 +67,12 @@ def main():
         help="Plot the predicted activity labels",
     )
     parser.add_argument(
-        "--cache-ssl",
+        "--cache-classifier",
         action="store_true",
-        help="Download and cache ssl module for offline usage",
+        help="Download and cache classifier file and model modules for offline usage",
+    )
+    parser.add_argument(
+        "--model-repo-path", "-m", help="Enter repository of ssl model", default=None
     )
     parser.add_argument("--quiet", "-q", action="store_true", help="Suppress output")
     args = parser.parse_args()
@@ -74,8 +81,18 @@ def main():
 
     verbose = not args.quiet
 
-    if args.cache_ssl:
-        model = ActivityClassifier(weights_path=True, ssl_repo=None, verbose=verbose)
+    classifier_path = (
+        pathlib.Path(__file__).parent / f"{__classifier_version__}.joblib.lzma"
+    )
+
+    if args.cache_classifier:
+        load_classifier(
+            classifier_path=classifier_path,
+            model_repo_path=None,
+            check_md5=True,
+            force_download=True,
+            verbose=verbose,
+        )
 
         after = time.time()
         print(f"Done! ({round(after - before,2)}s)")
@@ -95,21 +112,21 @@ def main():
     outdir = os.path.join(args.outdir, basename)
     os.makedirs(outdir, exist_ok=True)
 
-    # Run model
+    # Run classifier
     if verbose:
-        print("Loading model...")
-    model_path = pathlib.Path(__file__).parent / f"{__model_version__}.joblib.lzma"
-    check_md5 = args.model_path is None
-    model: ActivityClassifier = load_model(
-        args.model_path or model_path, check_md5, args.force_download
+        print("Loading classifier...")
+
+    check_md5 = args.classifier_path is None
+    classifier: ActivityClassifier = load_classifier(
+        args.classifier_path or classifier_path, args.model_repo_path, check_md5, args.force_download, verbose
     )
 
-    model.verbose = verbose
-    model.device = args.pytorch_device
+    classifier.verbose = verbose
+    classifier.device = args.pytorch_device
 
     if verbose:
         print("Running activity classifier...")
-    Y = model.predict_from_frame(data)
+    Y = classifier.predict_from_frame(data)
 
     # Save predicted activities
     timeSeriesFile = f"{outdir}/{basename}-timeSeries.csv.gz"
@@ -118,8 +135,17 @@ def main():
     if verbose:
         print("Time series output written to:", timeSeriesFile)
 
+    # Plot activity profile
+    if args.plot_activity:
+        plotFile = f"{outdir}/{basename}-timeSeries-plot.png"
+        fig = plotTimeSeries(Y)
+        fig.savefig(plotFile, dpi=200, bbox_inches="tight")
+
+        if verbose:
+            print("Output plot written to:", plotFile)
+
     # Summary
-    summary = getActivitySummary(Y, True, True, verbose)
+    summary = getActivitySummary(Y, classifier.labels, True, True, verbose)
 
     # Join the actipy processing info, with acitivity summary data
     outputSummary = {**summary, **info}
@@ -136,7 +162,7 @@ def main():
     if verbose:
         print("\nSummary Stats\n---------------------")
         print(
-            {
+            json.dumps({
                 key: outputSummary[key]
                 for key in [
                     "Filename",
@@ -144,21 +170,11 @@ def main():
                     "WearTime(days)",
                     "NonwearTime(days)",
                     "ReadOK",
-                    "acc-overall-avg(mg)",
                 ]
-                + [f"{label}-week-avg" for label in ACTIVITY_LABELS]
-            }
+                + [f"{label}-overall-avg" for label in ["acc"] + classifier.labels]
+            }, indent=4, cls=NpEncoder)
         )
 
-    # Plot activity profile
-    if args.plot_activity:
-        plotFile = f"{outdir}/{basename}-timeSeries-plot.png"
-        fig = plotTimeSeries(Y)
-        fig.savefig(plotFile, dpi=200, bbox_inches="tight")
-
-        if verbose:
-            print("Output plot written to:", plotFile)
-
     after = time.time()
     print(f"Done! ({round(after - before,2)}s)")
 
@@ -237,27 +253,41 @@ def resolve_path(path):
     return dirname, filename, extension
 
 
-def load_model(model_path, check_md5=True, force_download=False):
-    """Load trained model. Download if not exists."""
+def load_classifier(
+    classifier_path,
+    model_repo_path=None,
+    check_md5=True,
+    force_download=False,
+    verbose=True,
+):
+    """Load trained classifier. Download if not exists."""
 
-    pth = pathlib.Path(model_path)
+    pth = pathlib.Path(classifier_path)
 
     if force_download or not pth.exists():
 
-        url = f"{BASE_URL}{__model_version__}.joblib.lzma"
+        url = f"{BASE_URL}{__classifier_version__}.joblib.lzma"
 
-        print(f"Downloading {url}...")
+        if verbose:
+            print(f"Downloading {url}...")
 
         with urllib.request.urlopen(url) as f_src, open(pth, "wb") as f_dst:
             shutil.copyfileobj(f_src, f_dst)
 
     if check_md5:
-        assert md5(pth) == __model_md5__, (
-            "Model file is corrupted. Please run with --force-download "
+        assert md5(pth) == __classifier_md5__, (
+            "Classifier file is corrupted. Please run with --force-download "
             "to download the model file again."
         )
 
-    return joblib.load(pth)
+    classifier: ActivityClassifier = joblib.load(pth)
+
+    if model_repo_path and pathlib.Path(model_repo_path).exists() and verbose:
+        print(f"Loading model repository from {model_repo_path}.")
+
+    classifier.load_model(model_repo_path)
+
+    return classifier
 
 
 def md5(fname):

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/models.py

@@ -1,5 +1,6 @@
 import numpy as np
 import pandas as pd
+import joblib
 from tqdm.auto import tqdm
 from torch.utils.data import DataLoader
 
@@ -13,24 +14,24 @@ class ActivityClassifier:
         device="cpu",
         batch_size=512,
         window_sec=30,
-        weights_path="state_dict.pt",
+        weights_path=None,
         labels=[],
-        ssl_repo=None,
         repo_tag="v1.0.0",
         hmm_params=None,
         verbose=False,
     ):
         self.device = device
-        self.weights_path = weights_path
         self.repo_tag = repo_tag
         self.batch_size = batch_size
         self.window_sec = window_sec
-        self.state_dict = None
         self.labels = labels
         self.window_len = int(np.ceil(self.window_sec * sslmodel.SAMPLE_RATE))
         self.verbose = verbose
 
-        self.model = self._load_ssl(ssl_repo, weights_path)
+        self.model_weights = (
+            sslmodel.get_model_dict(weights_path, device) if weights_path else None
+        )
+        self.model = None
 
         hmm_params = hmm_params or dict()
         self.hmms = hmm.HMM(**hmm_params)
@@ -43,7 +44,7 @@ class ActivityClassifier:
             "batch_size: {self.batch_size}\n"
             "device: {self.device}\n"
             "hmm: {self.hmms}\n"
-            "model: {self.model}".format(self=self)
+            "model: {model}".format(self=self, model=self.model or "Model has not been loaded.")
         )
 
     def predict_from_frame(self, data):
@@ -72,6 +73,9 @@ class ActivityClassifier:
         return Y
 
     def _predict(self, X):
+        if self.model is None:
+            raise Exception("Model has not been loaded for ActivityClassifier.")
+
         sslmodel.verbose = self.verbose
 
         dataset = sslmodel.NormalDataset(X)
@@ -90,18 +94,18 @@ class ActivityClassifier:
 
         return y_pred
 
-    def _load_ssl(self, ssl_repo, weights):
-        model = sslmodel.get_sslnet(
-            self.device,
+    def load_model(self, model_repo=None):
+        self.model = sslmodel.get_sslnet(
             tag=self.repo_tag,
-            local_repo_path=ssl_repo,
-            pretrained=weights,
+            local_repo_path=model_repo,
+            pretrained_weights=self.model_weights or True,
             window_sec=self.window_sec,
             num_labels=len(self.labels),
         )
-        model.to(self.device)
+        self.model.to(self.device)
 
-        return model
+    def save(self, output_path):
+        joblib.dump(self, output_path, compress=("lzma", 3))
 
 
 def make_windows(data, window_sec, fn=None, return_index=False, verbose=True):

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/sslmodel.py

@@ -1,5 +1,6 @@
 """ Helper classes and functions for the SSL model """
 
+from collections import OrderedDict
 import torch
 import torch.nn as nn
 import numpy as np
@@ -184,10 +185,9 @@ class EarlyStopping:
 
 
 def get_sslnet(
-    device,
     tag="v1.0.0",
     local_repo_path=None,
-    pretrained=False,
+    pretrained_weights=False,
     window_sec: int = 30,
     num_labels: int = 4,
 ):
@@ -196,8 +196,8 @@ def get_sslnet(
 
     :param str device: PyTorch device to use
     :param str tag: Tag on the ssl-wearables repo to check out
-    :param str local_repo_path: Path to local version of the SSL repo for offline usage
-    :param bool/str pretrained: Initialise the model with UKB self-supervised pretrained weights
+    :param str local_repo_path: Path to local version of the SSL repository for offline usage
+    :param bool/OrderedDict pretrained_weights: Initialise the model with UKB self-supervised/specified pretrained weights
     :param int window_sec: The length of the window of data in seconds (limited to 5, 10 or 30)
     :param int num_labels: The number of labels to predict
     :return: pytorch SSL model
@@ -218,7 +218,7 @@ def get_sslnet(
             f"harnet{window_sec}",
             source="local",
             class_num=num_labels,
-            pretrained=pretrained == True,
+            pretrained=pretrained_weights == True,
         )
 
     else:
@@ -251,16 +251,20 @@ def get_sslnet(
             trust_repo=True,
             source=source,
             class_num=num_labels,
-            pretrained=pretrained == True,
+            pretrained=pretrained_weights == True,
             verbose=verbose,
         )
 
-    model_dict = torch.load(pretrained, map_location=device)
-    sslnet.load_state_dict(model_dict)
+    if isinstance(pretrained_weights, OrderedDict):
+        sslnet.load_state_dict(pretrained_weights)
 
     return sslnet
 
 
+def get_model_dict(weights_path, device):
+    return torch.load(weights_path, map_location=device)
+
+
 def predict(model, dataloader, device, output_logits=False):
     """
     Iterate over the dataloader and do prediction with a pytorch model.

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet/summarisation.py

@@ -7,11 +7,10 @@ from pandas.tseries.frequencies import to_offset
 from actinet.utils.utils import date_parser, toScreen
 from actinet import circadian
 
-ACTIVITY_LABELS = ["light", "moderate-vigorous", "sedentary", "sleep"]
-
 
 def getActivitySummary(
     data,
+    labels,
     intensityDistribution=False,
     circadianMetrics=False,
     verbose=True,
@@ -19,12 +18,12 @@ def getActivitySummary(
     """
     Calculate overall activity summary from predicted activity label data.
     This is achieved by:
-    1) check if data occurs at a daylight savings crossover
-    2) calculate imputation values to replace nan PA metric values
-    3) calculate empirical cumulative distribution function of vector magnitudes
-    4) derive main movement summaries (overall, weekday/weekend, and hour)
+    1) calculate imputation values to replace nan PA metric values
+    2) calculate empirical cumulative distribution function of vector magnitudes
+    3) derive main movement summaries (overall, weekday/weekend, and hour)
 
     :param str data: Input csv.gz file or pandas dataframe of processed epoch data
+    :param list(str) labels: Activity state labels
     :param bool intensityDistribution: Add intensity outputs to dict <summary>
     :param bool circadianMetrics: Add circadian rhythm metrics to dict <summary>
     :param bool verbose: Print verbose output
@@ -46,7 +45,7 @@ def getActivitySummary(
 
     # Main movement summaries
     summary = _summarise(
-        data, ACTIVITY_LABELS, intensityDistribution, circadianMetrics, verbose, summary
+        data, labels, intensityDistribution, circadianMetrics, verbose,
     )
 
     # Return physical activity summary
@@ -84,9 +83,6 @@ def _summarise(
     startTime = data.index[0]
     summary["FirstDay(0=mon,6=sun)"] = startTime.weekday()
 
-    # Check daylight savings crossings
-    summary = checkDST(data, summary)
-
     # Hours of activity for each recorded day
     epochPeriod = int(pd.Timedelta(freq).total_seconds())
     cols = labels
@@ -171,31 +167,22 @@ def _summarise(
     if circadianMetrics:
         toScreen("=== Calculating circadian metrics ===", verbose)
         summary = circadian.calculatePSD(
-            data, epochPeriod, False, ACTIVITY_LABELS, summary
+            data, epochPeriod, False, labels, summary
         )
         summary = circadian.calculatePSD(
-            data, epochPeriod, True, ACTIVITY_LABELS, summary
+            data, epochPeriod, True, labels, summary
         )
         summary = circadian.calculateFourierFreq(
-            data, epochPeriod, False, ACTIVITY_LABELS, summary
+            data, epochPeriod, False, labels, summary
         )
         summary = circadian.calculateFourierFreq(
-            data, epochPeriod, False, ACTIVITY_LABELS, summary
+            data, epochPeriod, False, labels, summary
         )
         summary = circadian.calculateM10L5(data, epochPeriod, summary)
 
     return summary
 
 
-def checkDST(data, summary={}):
-    if data.index[0].dst() < data.index[-1].dst():
-        summary["quality-daylightSavingsCrossover"] = 1
-    elif data.index[0].dst() > data.index[-1].dst():
-        summary["quality-daylightSavingsCrossover"] = -1
-    else:
-        summary["quality-daylightSavingsCrossover"] = 0
-
-    return summary
 
 
 def imputeMissing(data, extrapolate=True):
@@ -223,7 +210,7 @@ def imputeMissing(data, extrapolate=True):
                 data.index[0].floor("D"),
                 data.index[-1].ceil("D"),
                 freq=to_offset(pd.infer_freq(data.index)),
-                closed="left",
+                inclusive="left",
                 name="time",
             ),
             method="nearest",
@@ -303,7 +290,7 @@ def calculateECDF(x, summary):
     )
 
     # Write to summary
-    for level, val in ecdf.iteritems():
+    for level, val in ecdf.items():
         summary[f"{x.name}-ecdf-{level}mg"] = val
 
     return summary

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: actinet
-Version: 0.0.dev3
+Version: 0.0.dev5
 Summary: Activity detection algorithm compatible with the UK Biobank Accelerometer Dataset
 Home-page: https://github.com/OxWearables/actinet
 Download-URL: https://github.com/OxWearables/actinet
@@ -60,16 +60,16 @@ You are all set! The next time that you want to use `actinet`, open the Anaconda
 
 ```bash
 # Process an AX3 file
-$ actinet sample.cwa
+$ actinet -f sample.cwa
 
 # Or an ActiGraph file
-$ actinet sample.gt3x
+$ actinet -f sample.gt3x
 
 # Or a GENEActiv file
-$ actinet sample.bin
+$ actinet -f sample.bin
 
 # Or a CSV file (see data format below)
-$ actinet sample.csv
+$ actinet -f sample.csv
 ```
 
 ### Troubleshooting
@@ -80,12 +80,34 @@ Some systems may face issues with Java when running the script. If this is your
 conda install -n actinet openjdk=8
 ```
 
+### Offline usage
+
+To use this package offline, one must first download and install the relevant classifier file and model modules.
+This repository offers two ways of doing this.
+
+Run the following code when you have internet access:
+```console
+actinet --cache-classifier
+```
+ 
+Following this, the actinet classifier can be used as standard without internet access, without needing to specify the flags relating to the model repository.
+ 
+Alternatively, you can download or git clone the ssl modules from the [ssl-wearables repository](https://github.com/OxWearables/ssl-wearables).
+
+In addition, you can donwload/prepare a custom classifier file.
+
+Once this is downloaded to an appopriate location, you can run the actinet model using:
+ 
+```console
+actinet -f sample.cwa -c /path/to/classifier.joblib.lzma -m /path/to/ssl-wearables
+```
+
 ### Output files
 
 By default, output files will be stored in a folder named after the input file, `outputs/{filename}/`, created in the current working directory. You can change the output path with the `-o` flag:
 
 ```console
-$ actinet sample.cwa -o /path/to/some/folder/
+$ actinet -f sample.cwa -o /path/to/some/folder/
 
 <Output summary written to: /path/to/some/folder/sample-outputSummary.json>
 <Time series output written to: /path/to/some/folder/sample-timeSeries.csv.gz>
@@ -103,7 +125,7 @@ See [Data Dictionary](https://biobankaccanalysis.readthedocs.io/en/latest/datadi
 To plot the activity profiles, you can use the -p flag:
 
 ```console
-$ actinet sample.cwa -p
+$ actinet -f sample.cwa -p
 <Output plot written to: data/sample-timeSeries-plot.png>
 ```
 
@@ -138,9 +160,9 @@ To process multiple files you can create a text file in Notepad which includes o
 Example text file *commands.txt*:
 
 ```console
-actinet file1.cwa &
-actinet file2.cwa &
-actinet file3.cwa 
+actinet -f file1.cwa &
+actinet -f file2.cwa &
+actinet -f file3.cwa 
 :END
 ````
 
@@ -151,9 +173,9 @@ Once this file is created, run `cmd < commands.txt` from the terminal.
 Create a file *command.sh* with:
 
 ```console
-actinet file1.cwa
-actinet file2.cwa
-actinet file3.cwa
+actinet -f file1.cwa
+actinet -f file2.cwa
+actinet -f file3.cwa
 ```
 
 Then, run `bash command.sh` from the terminal.

{actinet-0.0.dev3 → actinet-0.0.dev5}/src/actinet.egg-info/requires.txt

@@ -3,7 +3,7 @@ numpy==1.24.*
 scipy==1.10.*
 pandas==2.0.*
 tqdm==4.64.*
-matplotlib==3.1.*
+matplotlib==3.5.*
 joblib==1.2.*
 scikit-learn==1.1.1
 imbalanced-learn==0.9.1