def render_top_k_summary(source_path: str, prefix: str, epoch_index: int):
""" Show top-k summary plot.
Args:
source_path (str): The root path of the dumped snapshots data for the corresponding experiment.
prefix (str): Prefix of data folders.
epoch_index (int): The index of selected epoch.
"""
helper.render_h3_title("Cike Bike Top K")
data = helper.read_detail_csv(
os.path.join(
source_path,
f"{prefix}{epoch_index}",
GlobalFileNames.stations_sum
)
)
# Convert index to station name.
index_name_conversion = helper.read_detail_csv(os.path.join(source_path, GlobalFileNames.name_convert))
data["station name"] = list(
map(
lambda x: index_name_conversion.loc[int(x[9:])][0],
data["name"]
)
)
# Generate top summary.
top_number = st.select_slider(
"Select Top K",
list(range(1, 6))
)
top_attributes = ["bikes", "trip_requirement", "fulfillment", "fulfillment_ratio"]
for item in top_attributes:
helper.generate_by_snapshot_top_summary("station name", data, int(top_number), item)
def _generate_top_k_summary(data: pd.DataFrame, snapshot_index: int, index_name_conversion: pd.DataFrame):
"""Generate CIM top k summary.
Args:
data (pd.Dataframe): Data of current snapshot.
snapshot_index (int): Selected snapshot index.
index_name_conversion (pd.Dataframe): Relationship between index and name.
"""
data_summary = data[data["frame_index"] == snapshot_index].reset_index(drop=True)
data_summary["fulfillment_ratio"] = list(
map(
lambda x, y: round(x / (y + 1 / 1000), 4),
data_summary["acc_fulfillment"],
data_summary["acc_booking"]
)
)
data_summary["port name"] = list(
map(
lambda x: index_name_conversion.loc[int(x[6:])][0],
data_summary["name"]
)
)
helper.render_h3_title("Select Top k:")
selected_top_number = st.select_slider(
label="",
options=list(range(1, 6))
)
top_attributes = CIMItemOption.acc_info + ["fulfillment_ratio"]
for item in top_attributes:
helper.generate_by_snapshot_top_summary(
"port name", data_summary, int(selected_top_number), item, snapshot_index
)
def _generate_intra_view_by_snapshot(data_stations: pd.DataFrame,
index_name_conversion: pd.DataFrame,
attribute_option_candidates: List[str],
snapshots_index: List[int],
snapshot_num: int, stations_num: int):
"""Show Citi Bike intra-view data by snapshot.
Args:
data_stations (pd.Dataframe): Filtered Data.
index_name_conversion (pd.Dataframe): Relationship between index and name.
attribute_option_candidates (List[str]): All options for users to choose.
snapshots_index (List[int]): Sampled snapshot index list.
snapshot_num (int): Number of snapshots.
stations_num (int): Number of stations.
"""
# Get selected snapshot index.
selected_snapshot = st.sidebar.select_slider(label="snapshot index",
options=snapshots_index)
helper.render_h3_title(f"Snapshot-{selected_snapshot}: Detail Data")
# Get according data with selected snapshot.
data_filtered = data_stations[data_stations["frame_index"] ==
selected_snapshot]
# Get increasing rate.
sample_ratio = helper.get_sample_ratio_selection_list(stations_num)
# Get sample rate (between 0-1).
selected_station_sample_ratio = st.sidebar.select_slider(
label="Station Sampling Ratio", options=sample_ratio)
# Get formula input and output.
data_formula = helper.get_filtered_formula_and_data(
GlobalScenarios.CITI_BIKE, data_filtered, attribute_option_candidates)
# Get sampled data and get station name.
down_pooling_sample_list = helper.get_sample_index_list(
stations_num, selected_station_sample_ratio)
attribute_option = data_formula["attribute_option"]
attribute_option.append("name")
if selected_station_sample_ratio == 0:
empty_head = attribute_option
empty_head.append("Station Name")
data_filtered = pd.DataFrame(columns=empty_head)
else:
data_filtered = data_formula["data"][attribute_option]
data_filtered = data_filtered.iloc[down_pooling_sample_list]
data_filtered["name"] = data_filtered["name"].apply(
lambda x: int(x[9:]))
data_filtered["Station Name"] = data_filtered["name"].apply(
lambda x: index_name_conversion.loc[int(x)])
data_melt = data_filtered.melt(["Station Name", "name"],
var_name="Attributes",
value_name="Count")
intra_snapshot_line_chart = alt.Chart(data_melt).mark_bar().encode(
x=alt.X("name:N", axis=alt.Axis(title="Name")),
y="Count:Q",
color="Attributes:N",
tooltip=["Attributes", "Count", "Station Name"],
).properties(width=700, height=380)
st.altair_chart(intra_snapshot_line_chart)
def _generate_intra_view_by_station(data_stations: pd.DataFrame,
index_name_conversion: pd.DataFrame,
attribute_option_candidates: List[str],
stations_index: List[int],
snapshot_num: int):
""" Show Citi Bike intra-view data by station.
Args:
data_stations (pd.Dataframe): Filtered station data.
index_name_conversion (pd.Dataframe): Relationship between index and name.
attribute_option_candidates (List[str]): All options for users to choose.
stations_index (List[int]): List of station index.
snapshot_num (int): Number of snapshots.
"""
selected_station = st.sidebar.select_slider(label="station index",
options=stations_index)
helper.render_h3_title(
index_name_conversion.loc[int(selected_station)][0] + " Detail Data")
# Filter data by station index.
data_filtered = data_stations[data_stations["name"] ==
f"stations_{selected_station}"]
snapshot_sample_ratio_list = helper.get_sample_ratio_selection_list(
snapshot_num)
selected_snapshot_sample_ratio = st.sidebar.select_slider(
label="Snapshot Sampling Ratio:", options=snapshot_sample_ratio_list)
# Get formula input and output.
data_formula = helper.get_filtered_formula_and_data(
GlobalScenarios.CITI_BIKE, data_filtered, attribute_option_candidates)
attribute_option = data_formula["attribute_option"]
attribute_option.append("frame_index")
data_filtered = data_formula["data"][attribute_option].reset_index(
drop=True)
down_pooling_sample_list = helper.get_sample_index_list(
snapshot_num, selected_snapshot_sample_ratio)
data_filtered = data_filtered.iloc[down_pooling_sample_list]
data_filtered.rename(columns={"frame_index": "snapshot_index"},
inplace=True)
data_melt = data_filtered.melt("snapshot_index",
var_name="Attributes",
value_name="Count")
intra_station_line_chart = alt.Chart(data_melt).mark_line().encode(
x=alt.X("snapshot_index", axis=alt.Axis(title="Snapshot Index")),
y="Count",
color="Attributes",
tooltip=["Attributes", "Count",
"snapshot_index"]).properties(width=700, height=380)
st.altair_chart(intra_station_line_chart)
intra_station_bar_chart = alt.Chart(data_melt).mark_bar().encode(
x=alt.X("snapshot_index:N", axis=alt.Axis(title="Snapshot Index")),
y="Count:Q",
color="Attributes:N",
tooltip=["Attributes", "Count",
"snapshot_index"]).properties(width=700, height=380)
st.altair_chart(intra_station_bar_chart)
def _render_intra_view_by_snapshot(source_path: str, option_epoch: int,
data_ports: pd.DataFrame,
snapshots_index: List[int],
index_name_conversion: pd.DataFrame,
attribute_option_candidates: List[str],
ports_num: int, prefix: str):
""" Show intra-view data by snapshot.
Args:
source_path (str): The root path of the dumped snapshots data for the corresponding experiment.
option_epoch (int): Index of selected epoch.
data_ports (pd.Dataframe): Filtered port data.
snapshots_index (List[int]): Index of selected snapshot.
index_name_conversion (pd.Dataframe): Relationship between index and name.
attribute_option_candidates (List[str]): All options for users to choose.
ports_num (int): Number of ports in current snapshot.
prefix (str): Prefix of data folders.
"""
selected_snapshot = st.sidebar.select_slider(label="snapshot index",
options=snapshots_index)
# Get sample ratio.
sample_ratio = helper.get_sample_ratio_selection_list(ports_num)
selected_port_sample_ratio = st.sidebar.select_slider(
label="Ports Sampling Ratio:", options=sample_ratio)
# Accumulated data.
helper.render_h1_title("Accumulated Data")
_render_intra_heat_map(source_path, GlobalScenarios.CIM, option_epoch,
selected_snapshot, prefix)
helper.render_h3_title(
f"SnapShot-{selected_snapshot}: Port Accumulated Attributes")
_generate_intra_panel_accumulated_by_snapshot(data_ports,
selected_snapshot, ports_num,
index_name_conversion,
selected_port_sample_ratio)
_generate_top_k_summary(data_ports, selected_snapshot,
index_name_conversion)
# Detailed data.
helper.render_h1_title("Detail Data")
_render_intra_panel_vessel(source_path, prefix, option_epoch,
selected_snapshot)
helper.render_h3_title(
f"Snapshot-{selected_snapshot}: Port Detail Attributes")
data_formula = helper.get_filtered_formula_and_data(
GlobalScenarios.CIM, data_ports, attribute_option_candidates)
_generate_intra_panel_by_snapshot(data_formula["data"], selected_snapshot,
ports_num, index_name_conversion,
selected_port_sample_ratio,
data_formula["attribute_option"])
def _render_intra_heat_map(source_path: str, scenario: GlobalScenarios,
epoch_index: int, snapshot_index: int, prefix: str):
"""Get matrix data and provide entrance to heat map of different scenario.
Args:
source_path (str): The root path of the dumped snapshots data for the corresponding experiment.
scenario (GlobalScenarios): Name of current scenario: CIM.
epoch_index (int): Selected epoch index.
snapshot_index (int): Selected snapshot index.
prefix (str): Prefix of data folders.
"""
matrix_data = pd.read_csv(
os.path.join(source_path, f"{prefix}{epoch_index}",
"matrices.csv")).loc[snapshot_index]
if scenario == GlobalScenarios.CIM:
helper.render_h3_title(
f"snapshot_{snapshot_index}: Accumulated Port Transfer Volume")
_generate_intra_heat_map(matrix_data["full_on_ports"])
def _generate_intra_panel_vessel(data_vessel: pd.DataFrame, snapshot_index: int, vessels_num: int):
"""Generate vessel data plot.
Args:
data_vessel (pd.Dataframe): Data of vessel information within selected snapshot index.
snapshot_index (int): User-selected snapshot index.
vessels_num (int): Number of vessels.
"""
helper.render_h3_title(f"SnapShot-{snapshot_index}: Vessel Attributes")
# Get sampled(and down pooling) index.
sample_ratio = helper.get_sample_ratio_selection_list(vessels_num)
selected_vessel_sample_ratio = st.sidebar.select_slider(
label="Vessels Sampling Ratio:",
options=sample_ratio
)
down_pooling_sample_list = helper.get_sample_index_list(vessels_num, selected_vessel_sample_ratio)
data_vessel = data_vessel[
data_vessel["frame_index"] == snapshot_index
][CIMItemOption.vessel_info].reset_index(drop=True)
data_rename = pd.DataFrame(columns=CIMItemOption.vessel_info)
for index in down_pooling_sample_list:
data_rename = pd.concat(
[data_rename, data_vessel[data_vessel["name"] == f"vessels_{index}"]],
axis=0
)
data_filtered = data_rename.reset_index(drop=True)
data_filtered["name"] = data_filtered["name"].apply(lambda x: int(x[8:]))
data_melt = data_filtered.melt(
"name",
var_name="Attributes",
value_name="Count"
)
intra_vessel_bar_chart = alt.Chart(data_melt).mark_bar().encode(
x=alt.X("name:N", axis=alt.Axis(title="Vessel Index")),
y="Count:Q",
color="Attributes:N",
tooltip=["Attributes", "Count", "name"]
).properties(
width=700,
height=380
)
st.altair_chart(intra_vessel_bar_chart)
def _render_intra_view_by_ports(
data_ports: pd.DataFrame, ports_index: int,
index_name_conversion: pd.DataFrame, attribute_option_candidates: List[str], snapshot_num: int
):
""" Show intra-view data by ports.
Args:
data_ports (pd.Dataframe): Filtered port data.
ports_index (int):Index of port of current data.
index_name_conversion (pd.Dataframe): Relationship of index and name.
attribute_option_candidates (List[str]): All options for users to choose.
snapshot_num (int): Number of snapshots on a port.
"""
selected_port = st.sidebar.select_slider(
label="Choose a Port:",
options=ports_index
)
sample_ratio = helper.get_sample_ratio_selection_list(snapshot_num)
selected_snapshot_sample_ratio = st.sidebar.select_slider(
label="Snapshot Sampling Ratio:",
options=sample_ratio
)
# Accumulated data.
helper.render_h1_title("CIM Accumulated Data")
helper.render_h3_title(
f"Port Accumulated Attributes: {selected_port} - {index_name_conversion.loc[int(selected_port)][0]}"
)
_generate_intra_panel_accumulated_by_ports(
data_ports, f"ports_{selected_port}", snapshot_num, selected_snapshot_sample_ratio
)
# Detailed data.
helper.render_h1_title("CIM Detail Data")
data_formula = helper.get_filtered_formula_and_data(
GlobalScenarios.CIM, data_ports, attribute_option_candidates
)
helper.render_h3_title(
f"Port Detail Attributes: {selected_port} - {index_name_conversion.loc[int(selected_port)][0]}"
)
_generate_intra_panel_by_ports(
data_formula["data"], f"ports_{selected_port}",
snapshot_num, selected_snapshot_sample_ratio, data_formula["attribute_option"]
)