dbscan_fl()

The function dbscan_fl() is a UDF (user-defined function) that clusterizes a dataset using the DBSCAN algorithm.

Syntax

T | invoke dbscan_fl(features, cluster_col, epsilon, min_samples, metric, metric_params)

Parameters

• For detailed description of the parameters, see DBSCAN documentation
• For the list of metrics see distance computations

Function definition

You can define the function by either embedding its code as a query-defined function, or creating it as a stored function in your database, as follows:

Query-defined

Define the function using the following let statement. No permissions are required.

let dbscan_fl=(tbl:(*), features:dynamic, cluster_col:string, epsilon:double, min_samples:int=10,
                       metric:string='minkowski', metric_params:dynamic=dynamic({'p': 2}))
{
    let kwargs = bag_pack('features', features, 'cluster_col', cluster_col, 'epsilon', epsilon, 'min_samples', min_samples,
                          'metric', metric, 'metric_params', metric_params);
    let code = ```if 1:

        from sklearn.cluster import DBSCAN
        from sklearn.preprocessing import StandardScaler

        features = kargs["features"]
        cluster_col = kargs["cluster_col"]
        epsilon = kargs["epsilon"]
        min_samples = kargs["min_samples"]
        metric = kargs["metric"]
        metric_params = kargs["metric_params"]

        df1 = df[features]
        mat = df1.values
        
        # Scale the dataframe
        scaler = StandardScaler()
        mat = scaler.fit_transform(mat)

        # see https://docs.scipy.org/doc/scipy/reference/spatial.distance.html for the various distance metrics

        dbscan = DBSCAN(eps=epsilon, min_samples=min_samples, metric=metric, metric_params=metric_params) # 'minkowski', 'chebyshev'
        labels = dbscan.fit_predict(mat)

        result = df
        result[cluster_col] = labels
    ```;
    tbl
    | evaluate python(typeof(*),code, kwargs)
};
// Write your query to use the function here.

Stored

Define the stored function once using the following .create function. Database User permissions are required.

.create-or-alter function with (folder = "Packages\\ML", docstring = "DBSCAN clustering")
dbscan_fl(tbl:(*), features:dynamic, cluster_col:string, epsilon:double, min_samples:int=10,
                       metric:string='minkowski', metric_params:dynamic=dynamic({'p': 2}))
{
    let kwargs = bag_pack('features', features, 'cluster_col', cluster_col, 'epsilon', epsilon, 'min_samples', min_samples,
                          'metric', metric, 'metric_params', metric_params);
    let code = ```if 1:

        from sklearn.cluster import DBSCAN
        from sklearn.preprocessing import StandardScaler

        features = kargs["features"]
        cluster_col = kargs["cluster_col"]
        epsilon = kargs["epsilon"]
        min_samples = kargs["min_samples"]
        metric = kargs["metric"]
        metric_params = kargs["metric_params"]

        df1 = df[features]
        mat = df1.values
        
        # Scale the dataframe
        scaler = StandardScaler()
        mat = scaler.fit_transform(mat)

        # see https://docs.scipy.org/doc/scipy/reference/spatial.distance.html for the various distance metrics

        dbscan = DBSCAN(eps=epsilon, min_samples=min_samples, metric=metric, metric_params=metric_params) # 'minkowski', 'chebyshev'
        labels = dbscan.fit_predict(mat)

        result = df
        result[cluster_col] = labels
    ```;
    tbl
    | evaluate python(typeof(*),code, kwargs)
}

Example

The following example uses the invoke operator to run the function.

Clustering of artificial dataset with three clusters

Query-defined

To use a query-defined function, invoke it after the embedded function definition.

let dbscan_fl=(tbl:(*), features:dynamic, cluster_col:string, epsilon:double, min_samples:int=10,
                       metric:string='minkowski', metric_params:dynamic=dynamic({'p': 2}))
{
    let kwargs = bag_pack('features', features, 'cluster_col', cluster_col, 'epsilon', epsilon, 'min_samples', min_samples,
                          'metric', metric, 'metric_params', metric_params);
    let code = ```if 1:

        from sklearn.cluster import DBSCAN
        from sklearn.preprocessing import StandardScaler

        features = kargs["features"]
        cluster_col = kargs["cluster_col"]
        epsilon = kargs["epsilon"]
        min_samples = kargs["min_samples"]
        metric = kargs["metric"]
        metric_params = kargs["metric_params"]

        df1 = df[features]
        mat = df1.values
        
        # Scale the dataframe
        scaler = StandardScaler()
        mat = scaler.fit_transform(mat)

        # see https://docs.scipy.org/doc/scipy/reference/spatial.distance.html for the various distance metrics

        dbscan = DBSCAN(eps=epsilon, min_samples=min_samples, metric=metric, metric_params=metric_params) # 'minkowski', 'chebyshev'
        labels = dbscan.fit_predict(mat)

        result = df
        result[cluster_col] = labels
    ```;
    tbl
    | evaluate python(typeof(*),code, kwargs)
};
union 
(range x from 1 to 100 step 1 | extend x=rand()+3, y=rand()+2),
(range x from 101 to 200 step 1 | extend x=rand()+1, y=rand()+4),
(range x from 201 to 300 step 1 | extend x=rand()+2, y=rand()+6)
| extend cluster_id=int(null)
| invoke dbscan_fl(pack_array("x", "y"), "cluster_id", epsilon=0.6, min_samples=4, metric_params=dynamic({'p':2}))
| render scatterchart with(series=cluster_id)

Stored

union 
(range x from 1 to 100 step 1 | extend x=rand()+3, y=rand()+2),
(range x from 101 to 200 step 1 | extend x=rand()+1, y=rand()+4),
(range x from 201 to 300 step 1 | extend x=rand()+2, y=rand()+6)
| extend cluster_id=int(null)
| invoke dbscan_fl(pack_array("x", "y"), "cluster_id", epsilon=0.6, min_samples=4, metric_params=dynamic({'p':2}))
| render scatterchart with(series=cluster_id)