davanstrien HF staff commited on
Commit
e19490a
1 Parent(s): 195a3cd

chore: Update dataset growth plot to exclude current month

Browse files
Files changed (1) hide show
  1. app.py +12 -7
app.py CHANGED
@@ -48,10 +48,14 @@ def plot_datasets_growth(data, framework):
48
  df = pd.DataFrame(data)
49
  df["createdAt"] = pd.to_datetime(df["createdAt"])
50
  df["month"] = df["createdAt"].dt.to_period("M").astype(str)
 
 
 
 
 
51
  df_counts = df.groupby("month").size().reset_index(name="count")
52
  df_counts["cumulative_count"] = df_counts["count"].cumsum()
53
  df_counts["growth_rate"] = df_counts["count"].pct_change()
54
-
55
  fig = px.line(df_counts, x="month", y="cumulative_count", title="Dataset Growth")
56
  fig.update_layout(
57
  xaxis_title="Month",
@@ -67,11 +71,9 @@ def plot_datasets_growth(data, framework):
67
  title="", orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1
68
  ),
69
  )
70
-
71
  fig.add_scatter(
72
  x=df_counts["month"], y=df_counts["growth_rate"], name="Growth Rate", yaxis="y2"
73
  )
74
-
75
  fig.update_layout(
76
  title={
77
  "text": f"Dataset Growth for {framework} datasets",
@@ -93,7 +95,6 @@ def plot_datasets_growth(data, framework):
93
  )
94
  ],
95
  )
96
-
97
  return fig
98
 
99
 
@@ -105,13 +106,17 @@ def update_dashboard(framework):
105
 
106
 
107
  with gr.Blocks() as demo:
108
- gr.Markdown("# View the growth of dataset frameworks/tags on the Hub")
 
 
 
109
  framework = gr.Dropdown(
110
  choices=choices,
111
  allow_custom_value=True,
 
112
  )
113
- plot = gr.Plot()
114
- markdown = gr.Markdown()
115
  framework.change(update_dashboard, inputs=[framework], outputs=[plot, markdown])
116
 
117
  demo.launch()
 
48
  df = pd.DataFrame(data)
49
  df["createdAt"] = pd.to_datetime(df["createdAt"])
50
  df["month"] = df["createdAt"].dt.to_period("M").astype(str)
51
+
52
+ # Exclude the current month
53
+ current_month = pd.Period.now("M").strftime("%Y-%m")
54
+ df = df[df["month"] < current_month]
55
+
56
  df_counts = df.groupby("month").size().reset_index(name="count")
57
  df_counts["cumulative_count"] = df_counts["count"].cumsum()
58
  df_counts["growth_rate"] = df_counts["count"].pct_change()
 
59
  fig = px.line(df_counts, x="month", y="cumulative_count", title="Dataset Growth")
60
  fig.update_layout(
61
  xaxis_title="Month",
 
71
  title="", orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1
72
  ),
73
  )
 
74
  fig.add_scatter(
75
  x=df_counts["month"], y=df_counts["growth_rate"], name="Growth Rate", yaxis="y2"
76
  )
 
77
  fig.update_layout(
78
  title={
79
  "text": f"Dataset Growth for {framework} datasets",
 
95
  )
96
  ],
97
  )
 
98
  return fig
99
 
100
 
 
106
 
107
 
108
  with gr.Blocks() as demo:
109
+ gr.Markdown("# Dataset frameworks/tags on the Hub")
110
+ gr.Markdown(
111
+ "This dashboard displays the number of datasets per author and the growth of datasets over time for a given framework/tag."
112
+ )
113
  framework = gr.Dropdown(
114
  choices=choices,
115
  allow_custom_value=True,
116
+ label="Select a framework/tag",
117
  )
118
+ plot = gr.Plot(label="Growth of datasets over time")
119
+ markdown = gr.Markdown(label="summary")
120
  framework.change(update_dashboard, inputs=[framework], outputs=[plot, markdown])
121
 
122
  demo.launch()