Initial Commit

Former-commit-id: c501f0e

.deployignore

@@ -1,3 +1,4 @@
 # these apps are ignored for deployment
 dash-summarize
-dash-translate
+dash-translate
+dash-cuml-umap

apps/dash-cuml-umap/.gitignore

@@ -0,0 +1,2 @@
+*__pycache__*
+*.vscode*

apps/dash-cuml-umap/Procfile

@@ -0,0 +1 @@
+web: gunicorn app:server --workers 1

apps/dash-cuml-umap/README.md

@@ -0,0 +1,31 @@
+# Dash cuML UMAP
+
+A demo of RAPIDS.ai's cuML UMAP functionality.
+
+## Instructions
+
+To get started, first clone this repo:
+```
+git clone https://github.com/plotly/dash-sample-apps.git
+cd dash-sample-apps/apps/dash-cuml-umap
+```
+
+Create a conda env and install the requirements:
+```
+conda create --name dash-cuml-umap --file requirements.txt
+```
+
+You can now run the app:
+```
+python app.py
+```
+
+and visit http://127.0.0.1:8050/.
+
+
+## Note about exporting
+
+Since RAPIDS.ai only support a conda environment, `pip freeze` the requirements would not work. Instead, do:
+```
+conda list -e > requirements.txt
+```

apps/dash-cuml-umap/app.py

@@ -0,0 +1,114 @@
+import time
+
+import cudf
+import cuml
+import cupy as cp
+import dash
+import dash_html_components as html
+import dash_core_components as dcc
+import dash_bootstrap_components as dbc
+from dash.dependencies import Input, Output, State
+import plotly.express as px
+
+
+# Load CSV into a cudf
+path = 'creditcard.csv'
+gdf = cudf.read_csv(path)
+gdf.Time = gdf.Time / 3600
+gdf.loc[gdf.Amount > 500, 'Amount'] = 500
+
+# Define app
+app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
+server = app.server
+
+
+controls = dbc.Row(
+    [
+        dbc.Col(
+            dbc.FormGroup(
+                [
+                    dbc.Label("Time since start (h)"),
+                    dcc.RangeSlider(
+                        id="slider-hours",
+                        min=0, max=50, step=1,
+                        value=[20, 30],
+                        marks={i: str(i) for i in range(0, 51, 10)}
+                    ),
+                ]
+            ),
+            md=6
+        ),
+
+        dbc.Col(
+            dbc.FormGroup(
+                [
+                    dbc.Label("Transaction Amount ($)"),
+                    dcc.RangeSlider(
+                        id="slider-amount",
+                        min=0, max=500, step=5,
+                        value=[200, 300],
+                        marks={i: str(i) for i in range(0, 501, 100)}
+                    ),
+                ]
+            ),
+            md=6
+        )
+    ],
+)
+
+
+# Define Layout
+app.layout = dbc.Container(
+    fluid=True,
+    children=[
+        html.H1("Dash cuML UMAP Demo"),
+        html.Hr(),
+        dbc.Card(controls, body=True),
+        dcc.Graph(id="graph-umap", style={'height': '70vh', 'max-height': '90vw'}),
+        html.Div(id='output-info'),
+    ],
+    style={'max-width': '960px', 'margin': 'auto'}
+)
+
+
+@app.callback(
+    [Output('graph-umap', 'figure'),
+     Output('output-info', 'children')],
+    [Input('slider-amount', 'value'),
+     Input('slider-hours', 'value'),]
+)
+def update_graph(amt, hrs):
+    t0 = time.time()
+    # First, filter based on the slider values
+    time_mask = (gdf.Time >= hrs[0]) & (gdf.Time <= hrs[1])
+    amount_mask = (gdf.Amount >= amt[0]) & (gdf.Amount <= amt[1])
+    filt_df = gdf.loc[time_mask & amount_mask]
+
+    # Then, select the features and train a UMAP model with cuML
+    features = filt_df.loc[:, "V1": "V28"].values
+    reducer = cuml.UMAP()
+    embedding = reducer.fit_transform(features)
+
+    # Convert the embedding back to numpy
+    embedding = cp.asnumpy(embedding)
+    amount = cp.asnumpy(filt_df.Amount.values.round(2))
+
+    # Create a plotly.express scatter plot
+    fig = px.scatter(
+        x=embedding[:, 0], 
+        y=embedding[:, 1], 
+        color=amount, 
+        labels={'color': 'Amount ($)'},
+        title='UMAP projection of credit card transactions'
+    )
+
+    t1 = time.time()
+    out_msg = f"Projected {embedding.shape[0]} transactions in {t1-t0:.2f}s."
+    alert = dbc.Alert(out_msg, color="success", dismissable=True)
+
+    return fig, alert
+
+
+
+if __name__ == "__main__":
+    app.run_server(debug=True)