Update app.py
Browse files
app.py
CHANGED
@@ -86,6 +86,71 @@ if st.session_state.df is not None and st.session_state.show_preview:
|
|
86 |
st.subheader("π Dataset Preview")
|
87 |
st.dataframe(st.session_state.df.head())
|
88 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
89 |
# Function to create TXT file
|
90 |
def create_text_report_with_viz_temp(report, conclusion, visualizations):
|
91 |
content = f"### Analysis Report\n\n{report}\n\n### Visualizations\n"
|
@@ -112,6 +177,59 @@ def create_text_report_with_viz_temp(report, conclusion, visualizations):
|
|
112 |
return temp_txt.name
|
113 |
|
114 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
# Function to create PDF with report text and visualizations
|
116 |
def create_pdf_report_with_viz(report, conclusion, visualizations):
|
117 |
pdf = FPDF()
|
@@ -301,27 +419,11 @@ if st.session_state.df is not None:
|
|
301 |
st.markdown(report_result if report_result else "β οΈ No Report Generated.")
|
302 |
|
303 |
# Step 4: Generate Visualizations
|
304 |
-
visualizations = []
|
305 |
|
306 |
-
fig_salary = px.box(st.session_state.df, x="job_title", y="salary_in_usd",
|
307 |
-
title="Salary Distribution by Job Title")
|
308 |
-
visualizations.append(fig_salary)
|
309 |
-
|
310 |
-
fig_experience = px.bar(
|
311 |
-
st.session_state.df.groupby("experience_level")["salary_in_usd"].mean().reset_index(),
|
312 |
-
x="experience_level", y="salary_in_usd",
|
313 |
-
title="Average Salary by Experience Level"
|
314 |
-
)
|
315 |
-
visualizations.append(fig_experience)
|
316 |
-
|
317 |
-
fig_employment = px.box(st.session_state.df, x="employment_type", y="salary_in_usd",
|
318 |
-
title="Salary Distribution by Employment Type")
|
319 |
-
visualizations.append(fig_employment)
|
320 |
|
321 |
# Step 5: Insert Visual Insights
|
322 |
st.markdown("### Visual Insights")
|
323 |
-
|
324 |
-
st.plotly_chart(fig, use_container_width=True)
|
325 |
|
326 |
# Step 6: Display Concise Conclusion
|
327 |
#st.markdown("#### Conclusion")
|
@@ -355,5 +457,4 @@ else:
|
|
355 |
# Sidebar Reference
|
356 |
with st.sidebar:
|
357 |
st.header("π Reference:")
|
358 |
-
st.markdown("[SQL Agents w CrewAI & Llama 3 - Plaban Nayak](https://github.com/plaban1981/Agents/blob/main/SQL_Agents_with_CrewAI_and_Llama_3.ipynb)")
|
359 |
-
|
|
|
86 |
st.subheader("π Dataset Preview")
|
87 |
st.dataframe(st.session_state.df.head())
|
88 |
|
89 |
+
# Ask GPT-4o for Visualization Suggestions
|
90 |
+
def ask_gpt4o_for_visualization(query, df, llm):
|
91 |
+
columns = ', '.join(df.columns)
|
92 |
+
prompt = f"""
|
93 |
+
Analyze the query and suggest the best visualization.
|
94 |
+
Query: "{query}"
|
95 |
+
Available Columns: {columns}
|
96 |
+
Respond in this JSON format:
|
97 |
+
{{
|
98 |
+
"chart_type": "bar/box/line/scatter",
|
99 |
+
"x_axis": "column_name",
|
100 |
+
"y_axis": "column_name",
|
101 |
+
"group_by": "optional_column_name"
|
102 |
+
}}
|
103 |
+
"""
|
104 |
+
response = llm.generate(prompt)
|
105 |
+
try:
|
106 |
+
return json.loads(response)
|
107 |
+
except json.JSONDecodeError:
|
108 |
+
st.error("β οΈ GPT-4o failed to generate a valid suggestion.")
|
109 |
+
return None
|
110 |
+
|
111 |
+
# Dynamically generate Plotly visualizations based on GPT-4o suggestions
|
112 |
+
def generate_visualization(suggestion, df):
|
113 |
+
chart_type = suggestion.get("chart_type", "bar").lower()
|
114 |
+
x_axis = suggestion.get("x_axis")
|
115 |
+
y_axis = suggestion.get("y_axis")
|
116 |
+
group_by = suggestion.get("group_by")
|
117 |
+
|
118 |
+
# Ensure required inputs are available
|
119 |
+
if not x_axis or not y_axis:
|
120 |
+
st.warning("β οΈ GPT-4o did not provide enough information for the visualization.")
|
121 |
+
return None
|
122 |
+
|
123 |
+
# Dynamically select the Plotly function
|
124 |
+
plotly_function = getattr(px, chart_type, None)
|
125 |
+
|
126 |
+
# Handle unsupported chart types gracefully
|
127 |
+
if not plotly_function:
|
128 |
+
st.warning(f"β οΈ Unsupported chart type '{chart_type}' suggested by GPT-4o.")
|
129 |
+
return None
|
130 |
+
|
131 |
+
# Prepare dynamic parameters for Plotly function
|
132 |
+
plot_args = {
|
133 |
+
"data_frame": df,
|
134 |
+
"x": x_axis,
|
135 |
+
"y": y_axis,
|
136 |
+
}
|
137 |
+
if group_by:
|
138 |
+
plot_args["color"] = group_by
|
139 |
+
|
140 |
+
try:
|
141 |
+
# Generate the dynamic visualization
|
142 |
+
fig = plotly_function(**plot_args)
|
143 |
+
fig.update_layout(
|
144 |
+
title=f"{chart_type.title()} Plot of {y_axis.replace('_', ' ').title()} by {x_axis.replace('_', ' ').title()}",
|
145 |
+
xaxis_title=x_axis.replace('_', ' ').title(),
|
146 |
+
yaxis_title=y_axis.replace('_', ' ').title(),
|
147 |
+
)
|
148 |
+
return fig
|
149 |
+
|
150 |
+
except Exception as e:
|
151 |
+
st.error(f"β οΈ Failed to generate visualization: {e}")
|
152 |
+
return None
|
153 |
+
|
154 |
# Function to create TXT file
|
155 |
def create_text_report_with_viz_temp(report, conclusion, visualizations):
|
156 |
content = f"### Analysis Report\n\n{report}\n\n### Visualizations\n"
|
|
|
177 |
return temp_txt.name
|
178 |
|
179 |
|
180 |
+
def add_stats_to_figure(fig, df, y_axis, chart_type):
|
181 |
+
# Calculate statistics
|
182 |
+
min_val = df[y_axis].min()
|
183 |
+
max_val = df[y_axis].max()
|
184 |
+
avg_val = df[y_axis].mean()
|
185 |
+
median_val = df[y_axis].median()
|
186 |
+
std_dev_val = df[y_axis].std()
|
187 |
+
|
188 |
+
# Stats summary text
|
189 |
+
stats_text = (
|
190 |
+
f"π **Statistics**\n\n"
|
191 |
+
f"- **Min:** ${min_val:,.2f}\n"
|
192 |
+
f"- **Max:** ${max_val:,.2f}\n"
|
193 |
+
f"- **Average:** ${avg_val:,.2f}\n"
|
194 |
+
f"- **Median:** ${median_val:,.2f}\n"
|
195 |
+
f"- **Std Dev:** ${std_dev_val:,.2f}"
|
196 |
+
)
|
197 |
+
|
198 |
+
# Charts suitable for stats annotations
|
199 |
+
if chart_type in ["bar", "line", "scatter"]:
|
200 |
+
# Add annotation box
|
201 |
+
fig.add_annotation(
|
202 |
+
text=stats_text,
|
203 |
+
xref="paper", yref="paper",
|
204 |
+
x=1.05, y=1,
|
205 |
+
showarrow=False,
|
206 |
+
align="left",
|
207 |
+
font=dict(size=12, color="black"),
|
208 |
+
bordercolor="black",
|
209 |
+
borderwidth=1,
|
210 |
+
bgcolor="rgba(255, 255, 255, 0.8)"
|
211 |
+
)
|
212 |
+
|
213 |
+
# Add horizontal lines for min, median, avg, max
|
214 |
+
fig.add_hline(y=min_val, line_dash="dot", line_color="red", annotation_text="Min", annotation_position="bottom right")
|
215 |
+
fig.add_hline(y=median_val, line_dash="dash", line_color="orange", annotation_text="Median", annotation_position="top right")
|
216 |
+
fig.add_hline(y=avg_val, line_dash="dashdot", line_color="green", annotation_text="Avg", annotation_position="top right")
|
217 |
+
fig.add_hline(y=max_val, line_dash="dot", line_color="blue", annotation_text="Max", annotation_position="top right")
|
218 |
+
|
219 |
+
elif chart_type == "box":
|
220 |
+
# Box plots already show distribution (no extra stats needed)
|
221 |
+
pass
|
222 |
+
|
223 |
+
elif chart_type == "pie":
|
224 |
+
# Pie charts don't need statistical overlays
|
225 |
+
st.info("π Pie charts focus on proportions. No additional stats displayed.")
|
226 |
+
|
227 |
+
else:
|
228 |
+
st.warning(f"β οΈ No stats added for unsupported chart type: {chart_type}")
|
229 |
+
|
230 |
+
return fig
|
231 |
+
|
232 |
+
|
233 |
# Function to create PDF with report text and visualizations
|
234 |
def create_pdf_report_with_viz(report, conclusion, visualizations):
|
235 |
pdf = FPDF()
|
|
|
419 |
st.markdown(report_result if report_result else "β οΈ No Report Generated.")
|
420 |
|
421 |
# Step 4: Generate Visualizations
|
|
|
422 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
423 |
|
424 |
# Step 5: Insert Visual Insights
|
425 |
st.markdown("### Visual Insights")
|
426 |
+
|
|
|
427 |
|
428 |
# Step 6: Display Concise Conclusion
|
429 |
#st.markdown("#### Conclusion")
|
|
|
457 |
# Sidebar Reference
|
458 |
with st.sidebar:
|
459 |
st.header("π Reference:")
|
460 |
+
st.markdown("[SQL Agents w CrewAI & Llama 3 - Plaban Nayak](https://github.com/plaban1981/Agents/blob/main/SQL_Agents_with_CrewAI_and_Llama_3.ipynb)")
|
|