updating 2weeks avg roi graph

app.py

@@ -162,7 +162,15 @@ def load_all_data():
         repo_type="dataset",
     )
     df10 = pd.read_parquet(weekly_avg_roi_pearl_agents)
-    return df1, df2, df3, df4, df5, df6, df7, df8, df9, df10
+
+    # two_weeks_avg_roi_pearl_agents.parquet
+    two_weeks_avg_roi_pearl_agents = hf_hub_download(
+        repo_id="valory/Olas-predict-dataset",
+        filename="two_weeks_avg_roi_pearl_agents.parquet",
+        repo_type="dataset",
+    )
+    df11 = pd.read_parquet(two_weeks_avg_roi_pearl_agents)
+    return df1, df2, df3, df4, df5, df6, df7, df8, df9, df10, df11
 
 
 def prepare_data():
@@ -178,6 +186,7 @@ def prepare_data():
         daa_qs_df,
         daa_pearl_df,
         weekly_avg_roi_pearl_agents,
+        two_weeks_avg_roi_pearl_agents,
     ) = load_all_data()
     all_trades["creation_timestamp"] = all_trades["creation_timestamp"].dt.tz_convert(
         "UTC"
@@ -256,6 +265,7 @@ def prepare_data():
         daa_qs_df,
         daa_pearl_df,
         weekly_avg_roi_pearl_agents,
+        two_weeks_avg_roi_pearl_agents,
     )
 
 
@@ -270,6 +280,7 @@ def prepare_data():
     daa_qs_df,
     daa_pearl_df,
     weekly_avg_roi_pearl_agents,
+    two_weeks_avg_roi_pearl_agents,
 ) = prepare_data()
 
 retention_df = prepare_retention_dataset(
@@ -571,11 +582,10 @@ with demo:
             with gr.Row():
                 gr.Markdown("# 2-weeks rolling average ROI for Pearl agents")
             with gr.Row():
-                # pearl_rolling_avg_plot = plot_rolling_average_roi(
-                #     traders_data=traders_data,
-                #     pearl_agents=weekly_avg_roi_pearl_agents,
-                # )
-                print("WIP")
+                pearl_rolling_avg_plot = plot_rolling_average_roi(
+                    two_weeks_avg_roi_pearl_agents
+                )
+
             with gr.Row():
                 gr.Markdown("# Average weekly ROI for Pearl agents")
             with gr.Row():

tabs/agent_graphs.py

@@ -24,37 +24,20 @@ def plot_rolling_average_dune(
     )
 
 
-def plot_rolling_average_roi(
-    traders_data: pd.DataFrame, pearl_agents: pd.DataFrame
-) -> gr.Plot:
-    """Function to plot the rolling average of ROI for pearl agents"""
-    # Get the list of unique addresses from the daa_pearl_df
-    unique_addresses = pearl_agents["safe_address"].unique()
-    # Filter the weekly_roi_df to include only those addresses
-    filtered_traders_data = traders_data[
-        traders_data["trader_address"].isin(unique_addresses)
-    ].copy()
-    # create the date column
-    filtered_traders_data["creation_timestamp"] = pd.to_datetime(
-        filtered_traders_data["creation_timestamp"]
-    )
-    filtered_traders_data["creation_date"] = filtered_traders_data[
-        "creation_timestamp"
-    ].dt.date
+def plot_rolling_average_roi(two_weeks_avg_roi_pearl_agents: pd.DataFrame) -> gr.Plot:
+    """Function to plot the 2-weeks rolling average ROI for pearl agents"""
 
-    # Get the 2-week rolling average of ROI
-    rolling_avg_roi_df = get_twoweeks_rolling_average_roi(filtered_traders_data)
     print("Rolling average ROI DataFrame:")
-    print(rolling_avg_roi_df.head())
+    print(two_weeks_avg_roi_pearl_agents.head())
 
     fig2 = px.bar(
-        rolling_avg_roi_df,
+        two_weeks_avg_roi_pearl_agents,
         x="creation_date",
-        y="roi",
+        y="two_weeks_avg_roi",
     )
     fig2.update_layout(
         xaxis_title="Week",
-        yaxis_title="2-week rolling average of ROI at the trader level",
+        yaxis_title="2-week rolling average ROI of pearl agents",
     )
 
     return gr.Plot(
@@ -62,60 +45,6 @@ def plot_rolling_average_roi(
     )
 
 
-def get_twoweeks_rolling_average_roi(traders_data: pd.DataFrame) -> pd.DataFrame:
-    """Function to get the 2-week rolling average of the ROI by market_creator and total"""
-
-    # Create a copy to avoid SettingWithCopyWarning
-    local_df = traders_data.copy()
-
-    # Ensure creation_date is datetime64[ns]
-    # Since creation_date comes from .dt.date, it's a date object, not datetime
-    local_df["creation_date"] = pd.to_datetime(local_df["creation_date"])
-
-    # Aggregate ROI at the date level
-    daily_avg2 = local_df.groupby("creation_date")["roi"].mean().reset_index()
-
-    # Set the datetime index
-    daily_avg2 = daily_avg2.set_index("creation_date")
-
-    # Now resample and rolling average
-    weekly_avg2 = daily_avg2.resample("W").mean()
-    rolling_avg = weekly_avg2.rolling(window=2).mean().reset_index()
-
-    return rolling_avg
-
-
-def get_weekly_average_roi(traders_data: pd.DataFrame) -> pd.DataFrame:
-    """Function to get the weekly average ROI by market_creator and total"""
-
-    # Create a copy to avoid SettingWithCopyWarning
-    local_df = traders_data.copy()
-
-    # Ensure creation_date is datetime64[ns]
-    # Since creation_date comes from .dt.date, it's a date object, not datetime
-    local_df["creation_date"] = pd.to_datetime(local_df["creation_date"])
-    # take the daily mean roi at the trader_address level
-    daily_mean_roi = (
-        local_df.groupby(["trader_address", "creation_date"])["roi"]
-        .mean()
-        .reset_index()
-    )
-
-    # Aggregate ROI at the date level first
-    daily_avg = daily_mean_roi.groupby("creation_date")["roi"].mean().reset_index()
-
-    # Set the datetime index
-    daily_avg = daily_avg.set_index("creation_date")
-
-    # Resample to weekly frequency and calculate mean
-    weekly_avg = daily_avg.resample("W").mean().reset_index()
-
-    # Remove NaN values
-    weekly_avg = weekly_avg.dropna(subset=["roi"])
-
-    return weekly_avg
-
-
 def plot_weekly_average_roi(weekly_avg_roi_df: pd.DataFrame) -> gr.Plot:
     """Function to plot the weekly average of ROI for pearl agents"""