Filters zhaw

docs/api_reference.rst

@@ -23,6 +23,7 @@ to be applied on every :class:`~traffic.core.Flight`  in the collection
 and stack them before triggering only one iteration.
 
 .. toctree::
+   :maxdepth: 1
 
    api_reference/traffic.core.aero
    api_reference/traffic.core.airspace
@@ -39,3 +40,4 @@ and stack them before triggering only one iteration.
    api_reference/traffic.core.sv
    api_reference/traffic.core.time
    api_reference/traffic.core.traffic
+   api_reference/traffic.algorithms.filters

docs/api_reference/traffic.algorithms.filters.rst

@@ -0,0 +1,93 @@
+traffic.algorithms.filters
+==========================
+
+.. jupyter-execute::
+    :hide-output:
+    :hide-code:
+
+    import altair as alt
+    from traffic.core import Flight
+    from traffic.data.samples import noisy
+
+    raw_data = noisy.assign(type="raw data")
+    default = noisy.filter("default").assign(type="default")
+    aggressive = noisy.filter("aggressive").assign(type="aggressive")
+
+    domain = ["raw data", "default", "aggressive"]
+
+    features = [
+        "altitude (in ft)",
+        "track angle (in deg)",
+        "vertical_rate (in ft/min)",
+        "groundspeed (in kts)",
+    ]
+
+
+    def chart(filtered: "Flight", titles: list[str] = features) -> alt.Chart:
+        def y_channel(title: str):
+            return alt.Y(title.split()[0], scale=alt.Scale(zero=False), title=None)
+
+        color_channel = alt.Color(
+            "type",
+            scale=alt.Scale(domain=domain),
+            legend=alt.Legend(title="Filter type"),
+        )
+        reference = raw_data.chart()
+        compared = filtered.chart()
+        return (
+            alt.vconcat(
+                alt.concat(
+                    alt.layer(
+                        reference.encode(y_channel(titles[0]), color_channel),
+                        compared.encode(y_channel(titles[0]), color_channel),
+                    ).properties(width=300, height=150, title=titles[0]),
+                    alt.layer(
+                        reference.encode(y_channel(titles[1]), color_channel),
+                        compared.encode(y_channel(titles[1]), color_channel),
+                    ).properties(width=300, height=150, title=titles[1]),
+                ),
+                alt.concat(
+                    alt.layer(
+                        reference.encode(y_channel(titles[2]), color_channel),
+                        compared.encode(y_channel(titles[2]), color_channel),
+                    ).properties(width=300, height=150, title=titles[2]),
+                    alt.layer(
+                        reference.encode(y_channel(titles[3]), color_channel),
+                        compared.encode(y_channel(titles[3]), color_channel),
+                    ).properties(width=300, height=150, title=titles[3]),
+                ),
+            )
+            .configure_axisX(format="%H:%M", title=None, labelFontSize=12)
+            .configure_axisY(labelFontSize=12)
+            .configure_legend(orient="bottom", labelFontSize=14, titleFontSize=14)
+            .configure_title(anchor="start", fontSize=16)
+        )
+
+traffic comes with some pre-implemented filters to be passed to the
+:meth:`~traffic.core.Flight.filter` method. The method takes either a
+:class:`~traffic.algorithms.filters.FilterBase` instance, or a string parameter:
+
+- ``"default"`` is a relatively fast option with decent performance on
+  trajectories extracted from the OpenSky database (with their most common
+  glitches)
+
+.. jupyter-execute::
+    :hide-code:
+
+    chart(default)
+
+- ``"aggressive"`` is a composition of several filters which may result in
+  smoother trajectories.
+
+.. jupyter-execute::
+    :hide-code:
+
+    chart(aggressive)
+
+
+API reference
+-------------
+
+.. automodule:: traffic.algorithms.filters
+    :members:
+    :show-inheritance: