Real world exmpl

In [1]:

from river.datasets import Insects
from river.preprocessing import MinMaxScaler
from river.metrics import Accuracy
from river.utils import Rolling
from river.tree import HoeffdingTreeClassifier
from deep_river.classification import Classifier
from torch import nn
from tqdm import tqdm
import matplotlib.pyplot as plt
import matplotlib
import torch

from river.datasets import Insects from river.preprocessing import MinMaxScaler from river.metrics import Accuracy from river.utils import Rolling from river.tree import HoeffdingTreeClassifier from deep_river.classification import Classifier from torch import nn from tqdm import tqdm import matplotlib.pyplot as plt import matplotlib import torch

In [2]:

insects = Insects("abrupt_balanced")


def test_train_eval(model, stream, update_interval=100):
    results = []
    steps = []
    step = 0
    metric = Rolling(Accuracy(), window_size=400)
    scaler = MinMaxScaler()
    for x, y in tqdm(list(stream)):
        x = scaler.learn_one(x).transform_one(x)
        y_pred = model.predict_one(x)
        model.learn_one(x, y)
        metric.update(y, y_pred)
        step += 1
        if step % update_interval == 0:
            results.append(metric.get())
            steps.append(step)
    return steps, results

insects = Insects("abrupt_balanced") def test_train_eval(model, stream, update_interval=100): results = [] steps = [] step = 0 metric = Rolling(Accuracy(), window_size=400) scaler = MinMaxScaler() for x, y in tqdm(list(stream)): x = scaler.learn_one(x).transform_one(x) y_pred = model.predict_one(x) model.learn_one(x, y) metric.update(y, y_pred) step += 1 if step % update_interval == 0: results.append(metric.get()) steps.append(step) return steps, results

In [3]:

class SimpleMLP(nn.Module):
    def __init__(self, n_features):
        super().__init__()
        self.hidden1 = nn.Linear(n_features, 30)
        self.hidden2 = nn.Linear(30, 30)
        self.logits = nn.Linear(30, 6)

    def forward(self, x):
        h = torch.relu(self.hidden1(x))
        h = torch.relu(self.hidden2(h))
        return self.logits(h)


mlp = Classifier(
    SimpleMLP,
    loss_fn="binary_cross_entropy_with_logits",
    optimizer_fn="sgd",
    lr=0.05,
    seed=42,
)
steps, results_mlp = test_train_eval(mlp, insects)

class SimpleMLP(nn.Module): def __init__(self, n_features): super().__init__() self.hidden1 = nn.Linear(n_features, 30) self.hidden2 = nn.Linear(30, 30) self.logits = nn.Linear(30, 6) def forward(self, x): h = torch.relu(self.hidden1(x)) h = torch.relu(self.hidden2(h)) return self.logits(h) mlp = Classifier( SimpleMLP, loss_fn="binary_cross_entropy_with_logits", optimizer_fn="sgd", lr=0.05, seed=42, ) steps, results_mlp = test_train_eval(mlp, insects)

---------------------------------------------------------------------------
HTTPError                                 Traceback (most recent call last)
Cell In[3], line 21
     11         return self.logits(h)
     14 mlp = Classifier(
     15     SimpleMLP,
     16     loss_fn="binary_cross_entropy_with_logits",
   (...)
     19     seed=42,
     20 )
---> 21 steps, results_mlp = test_train_eval(mlp, insects)

Cell In[2], line 9, in test_train_eval(model, stream, update_interval)
      7 metric = Rolling(Accuracy(), window_size=400)
      8 scaler = MinMaxScaler()
----> 9 for x, y in tqdm(list(stream)):
     10     x = scaler.learn_one(x).transform_one(x)
     11     y_pred = model.predict_one(x)

File ~/Documents/Environments/deep-river/lib/python3.11/site-packages/river/datasets/base.py:325, in RemoteDataset.__iter__(self)
    323 def __iter__(self):
    324     if not self.is_downloaded:
--> 325         self.download(verbose=True)
    326     if not self.is_downloaded:
    327         raise RuntimeError("Something went wrong during the download")

File ~/Documents/Environments/deep-river/lib/python3.11/site-packages/river/datasets/base.py:272, in RemoteDataset.download(self, force, verbose)
    269 directory.mkdir(parents=True, exist_ok=True)
    270 archive_path = directory.joinpath(os.path.basename(self.url))
--> 272 with request.urlopen(self.url) as r:
    273     # Notify the user
    274     if verbose:
    275         meta = r.info()

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:216, in urlopen(url, data, timeout, cafile, capath, cadefault, context)
    214 else:
    215     opener = _opener
--> 216 return opener.open(url, data, timeout)

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:525, in OpenerDirector.open(self, fullurl, data, timeout)
    523 for processor in self.process_response.get(protocol, []):
    524     meth = getattr(processor, meth_name)
--> 525     response = meth(req, response)
    527 return response

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:634, in HTTPErrorProcessor.http_response(self, request, response)
    631 # According to RFC 2616, "2xx" code indicates that the client's
    632 # request was successfully received, understood, and accepted.
    633 if not (200 <= code < 300):
--> 634     response = self.parent.error(
    635         'http', request, response, code, msg, hdrs)
    637 return response

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:563, in OpenerDirector.error(self, proto, *args)
    561 if http_err:
    562     args = (dict, 'default', 'http_error_default') + orig_args
--> 563     return self._call_chain(*args)

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:496, in OpenerDirector._call_chain(self, chain, kind, meth_name, *args)
    494 for handler in handlers:
    495     func = getattr(handler, meth_name)
--> 496     result = func(*args)
    497     if result is not None:
    498         return result

File /usr/local/Cellar/python@3.11/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/urllib/request.py:643, in HTTPDefaultErrorHandler.http_error_default(self, req, fp, code, msg, hdrs)
    642 def http_error_default(self, req, fp, code, msg, hdrs):
--> 643     raise HTTPError(req.full_url, code, msg, hdrs, fp)

HTTPError: HTTP Error 404: Not Found

In [ ]:

tree = HoeffdingTreeClassifier()
steps, results_tree = test_train_eval(tree, insects)

tree = HoeffdingTreeClassifier() steps, results_tree = test_train_eval(tree, insects)

In [5]:

fig, ax = plt.subplots(figsize=(8, 4))
ax.plot(steps, results_mlp, label="MLP")
ax.plot(steps, results_tree, label="Hoeffding Tree")

change_points = [0, 14352, 19500, 33240, 38682, 39510, 52848]
temps = [30, 20, 36, 26, 34, 30]
cmap = plt.cm.plasma
norm = matplotlib.colors.Normalize(vmin=20, vmax=36)
sm = plt.cm.ScalarMappable(cmap=cmap, norm=norm)
for i in range(len(change_points) - 1):
    ax.axvspan(
        change_points[i],
        change_points[i + 1],
        alpha=0.4,
        color=cmap(norm(temps[i])),
        lw=0,
    )
ax.set_xlim(0, change_points[-1])
cbar = fig.colorbar(sm, ax=ax)
cbar.ax.set_ylabel("Temperature in °C")
ax.set_xlabel("Steps")
ax.set_ylabel("Moving Accuracy")
ax.legend()

fig, ax = plt.subplots(figsize=(8, 4)) ax.plot(steps, results_mlp, label="MLP") ax.plot(steps, results_tree, label="Hoeffding Tree") change_points = [0, 14352, 19500, 33240, 38682, 39510, 52848] temps = [30, 20, 36, 26, 34, 30] cmap = plt.cm.plasma norm = matplotlib.colors.Normalize(vmin=20, vmax=36) sm = plt.cm.ScalarMappable(cmap=cmap, norm=norm) for i in range(len(change_points) - 1): ax.axvspan( change_points[i], change_points[i + 1], alpha=0.4, color=cmap(norm(temps[i])), lw=0, ) ax.set_xlim(0, change_points[-1]) cbar = fig.colorbar(sm, ax=ax) cbar.ax.set_ylabel("Temperature in °C") ax.set_xlabel("Steps") ax.set_ylabel("Moving Accuracy") ax.legend()

Out[5]:

<matplotlib.legend.Legend at 0x142fbceb0>