Skip to content

Instantly share code, notes, and snippets.

@abulte

abulte/README.md

Created January 21, 2024 09:33
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save abulte/5a9989b0ce7132650241beef174a1a84 to your computer and use it in GitHub Desktop.
Save abulte/5a9989b0ce7132650241beef174a1a84 to your computer and use it in GitHub Desktop.
Download ZIP
pandas-ai with mistral llms
Raw
README.md

Testing pandas-ai with mistral hosted LLMs

# patched pandas-ai working with llm module
pip install -e . ../Developer/pandas-ai
# patched llm-mistral that supports `.key` to pass creds
pip install git+https://github.com/abulte/llm-mistral.git@patch-1
echo "export MISTRAL_API_KEY=xxx" > .envrc
direnv allow
Display the source blob
Display the rendered blob
Raw
pandas-ai.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "e48e17c6-3165-4c72-8eaf-d81edaeef392",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import pandas as pd\n",
"from pandasai import SmartDataframe\n",
"from pandasai.llm.llm import LlmLLM"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "4bc71685-9c92-4e75-b93e-90ccbb985253",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>day</th>\n",
" <th>start_local</th>\n",
" <th>start_utc</th>\n",
" <th>end_local</th>\n",
" <th>end_utc</th>\n",
" <th>reading_time_seconds</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>2017-09-15</td>\n",
" <td>NaN</td>\n",
" <td>NaT</td>\n",
" <td>2017-09-16 01:54:41+02:00</td>\n",
" <td>2017-09-15 23:54:41+00:00</td>\n",
" <td>NaN</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>2017-09-16</td>\n",
" <td>2017-09-16 18:54:12+02:00</td>\n",
" <td>2017-09-16 16:54:12+00:00</td>\n",
" <td>2017-09-17 00:21:55+02:00</td>\n",
" <td>2017-09-16 22:21:55+00:00</td>\n",
" <td>19663.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2017-09-17</td>\n",
" <td>NaN</td>\n",
" <td>NaT</td>\n",
" <td>2017-09-17 15:23:55+02:00</td>\n",
" <td>2017-09-17 13:23:55+00:00</td>\n",
" <td>NaN</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2017-10-29</td>\n",
" <td>2017-10-29 19:06:31+01:00</td>\n",
" <td>2017-10-29 18:06:31+00:00</td>\n",
" <td>2017-10-29 19:06:32+01:00</td>\n",
" <td>2017-10-29 18:06:32+00:00</td>\n",
" <td>1.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>2017-12-05</td>\n",
" <td>2017-12-05 22:52:00+01:00</td>\n",
" <td>2017-12-05 21:52:00+00:00</td>\n",
" <td>2017-12-05 23:02:46+01:00</td>\n",
" <td>2017-12-05 22:02:46+00:00</td>\n",
" <td>646.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2101</th>\n",
" <td>2024-01-12</td>\n",
" <td>2024-01-12 22:48:52+01:00</td>\n",
" <td>2024-01-12 21:48:52+00:00</td>\n",
" <td>2024-01-13 00:54:24+01:00</td>\n",
" <td>2024-01-12 23:54:24+00:00</td>\n",
" <td>7532.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2102</th>\n",
" <td>2024-01-13</td>\n",
" <td>2024-01-13 21:51:03+01:00</td>\n",
" <td>2024-01-13 20:51:03+00:00</td>\n",
" <td>2024-01-14 00:40:36+01:00</td>\n",
" <td>2024-01-13 23:40:36+00:00</td>\n",
" <td>10173.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2103</th>\n",
" <td>2024-01-14</td>\n",
" <td>2024-01-14 22:42:42+01:00</td>\n",
" <td>2024-01-14 21:42:42+00:00</td>\n",
" <td>2024-01-15 01:06:01+01:00</td>\n",
" <td>2024-01-15 00:06:01+00:00</td>\n",
" <td>8599.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2104</th>\n",
" <td>2024-01-15</td>\n",
" <td>2024-01-15 21:56:59+01:00</td>\n",
" <td>2024-01-15 20:56:59+00:00</td>\n",
" <td>2024-01-16 00:07:04+01:00</td>\n",
" <td>2024-01-15 23:07:04+00:00</td>\n",
" <td>7805.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2105</th>\n",
" <td>2024-01-16</td>\n",
" <td>2024-01-16 22:36:09+01:00</td>\n",
" <td>2024-01-16 21:36:09+00:00</td>\n",
" <td>2024-01-16 22:41:46+01:00</td>\n",
" <td>2024-01-16 21:41:46+00:00</td>\n",
" <td>337.0</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>2106 rows × 6 columns</p>\n",
"</div>"
],
"text/plain": [
" day start_local start_utc \\\n",
"0 2017-09-15 NaN NaT \n",
"1 2017-09-16 2017-09-16 18:54:12+02:00 2017-09-16 16:54:12+00:00 \n",
"2 2017-09-17 NaN NaT \n",
"3 2017-10-29 2017-10-29 19:06:31+01:00 2017-10-29 18:06:31+00:00 \n",
"4 2017-12-05 2017-12-05 22:52:00+01:00 2017-12-05 21:52:00+00:00 \n",
"... ... ... ... \n",
"2101 2024-01-12 2024-01-12 22:48:52+01:00 2024-01-12 21:48:52+00:00 \n",
"2102 2024-01-13 2024-01-13 21:51:03+01:00 2024-01-13 20:51:03+00:00 \n",
"2103 2024-01-14 2024-01-14 22:42:42+01:00 2024-01-14 21:42:42+00:00 \n",
"2104 2024-01-15 2024-01-15 21:56:59+01:00 2024-01-15 20:56:59+00:00 \n",
"2105 2024-01-16 2024-01-16 22:36:09+01:00 2024-01-16 21:36:09+00:00 \n",
"\n",
" end_local end_utc \\\n",
"0 2017-09-16 01:54:41+02:00 2017-09-15 23:54:41+00:00 \n",
"1 2017-09-17 00:21:55+02:00 2017-09-16 22:21:55+00:00 \n",
"2 2017-09-17 15:23:55+02:00 2017-09-17 13:23:55+00:00 \n",
"3 2017-10-29 19:06:32+01:00 2017-10-29 18:06:32+00:00 \n",
"4 2017-12-05 23:02:46+01:00 2017-12-05 22:02:46+00:00 \n",
"... ... ... \n",
"2101 2024-01-13 00:54:24+01:00 2024-01-12 23:54:24+00:00 \n",
"2102 2024-01-14 00:40:36+01:00 2024-01-13 23:40:36+00:00 \n",
"2103 2024-01-15 01:06:01+01:00 2024-01-15 00:06:01+00:00 \n",
"2104 2024-01-16 00:07:04+01:00 2024-01-15 23:07:04+00:00 \n",
"2105 2024-01-16 22:41:46+01:00 2024-01-16 21:41:46+00:00 \n",
"\n",
" reading_time_seconds \n",
"0 NaN \n",
"1 19663.0 \n",
"2 NaN \n",
"3 1.0 \n",
"4 646.0 \n",
"... ... \n",
"2101 7532.0 \n",
"2102 10173.0 \n",
"2103 8599.0 \n",
"2104 7805.0 \n",
"2105 337.0 \n",
"\n",
"[2106 rows x 6 columns]"
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df = pd.read_csv(f\"../output/2024-01-19.csv\", parse_dates=[\"day\", \"start_utc\", \"end_utc\"])\n",
"df"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "a655fcf8-de85-48c9-8a64-cd83eb2ce081",
"metadata": {},
"outputs": [],
"source": [
"llm = LlmLLM(api_key=os.getenv(\"MISTRAL_API_KEY\"))"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "937f6bc3-2f34-4071-ad1a-531aa3ec87df",
"metadata": {},
"outputs": [],
"source": [
"df = SmartDataframe(df, config={\"llm\": llm})"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "499dc433-be76-4407-8a95-149a8ab332ba",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Here is the updated code:\n",
"\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs = [pd.read_csv(pd.compat.StringIO(data), parse_dates=['day', 'start_local', 'start_utc', 'end_local', 'end_utc'] ) for data in df.split('\\n') if data]\n",
"\n",
"result = {\"type\": \"dataframe\", \"value\": dfs[0].columns}\n",
"```\n",
"\n",
"In this code, I first load the data from the provided string into a list of dataframes `dfs`. I then set the `result` variable to be a dictionary with a type of \"dataframe\" and the value being the columns of the first dataframe in the list.\n",
"The error is occurring because `dfs[0]` is a pandas Index object, not a DataFrame. This is happening because `pd.read_csv` returns a DataFrame with the column named 'day' as the index by default. To fix this, we can add `index_col=False` to the `pd.read_csv` call to prevent pandas from setting the 'day' column as the index.\n",
"\n",
"Here's the updated code:\n",
"\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs = [pd.read_csv(pd.compat.StringIO(data), parse_dates=['day', 'start_local', 'start_utc', 'end_local', 'end_utc'], index_col=False ) for data in df.split('\\n') if data]\n",
"\n",
"result = {\"type\": \"dataframe\", \"value\": dfs[0].columns}\n",
"```\n",
"\n",
"With this updated code, the `result` value will be a dictionary with the key \"value\" set to a list of the column names in the DataFrame `dfs[0]`.\n",
"The error is caused by trying to convert the index of the dataframe to JSON. To fix this, we can reset the index of the dataframe before converting it to a dictionary.\n",
"\n",
"Here is the corrected code:\n",
"\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs = [pd.read_csv(pd.compat.StringIO(data), parse_dates=['day', 'start_local', 'start_utc', 'end_local', 'end_utc'] ) for data in df.split('\\n') if data]\n",
"\n",
"# Reset the index of the first dataframe\n",
"dfs[0] = dfs[0].reset_index(drop=True)\n",
"\n",
"result = {\"type\": \"dataframe\", \"value\": dfs[0].columns}\n",
"```\n",
"\n",
"In this code, we reset the index of the first dataframe `dfs[0]` by calling `dfs[0].reset_index(drop=True)`. This creates a new dataframe with a reset index, which can be converted to a dictionary without errors.\n"
]
},
{
"data": {
"text/plain": [
"\"Unfortunately, I was not able to answer your question, because of the following error:\\n\\n'Index' object has no attribute 'to_json'\\n\""
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df.chat('show the columns')"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "62bafabb-b8f8-4d57-a9af-f0c2ed2b9ad8",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Sure, here is the updated code:\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs[0]['start_local'] = pd.to_datetime(dfs[0]['start_local'])\n",
"grouped = dfs[0].groupby(dfs[0]['start_local'].dt.month).mean()\n",
"grouped.reset_index(inplace=True)\n",
"grouped.columns = ['month', 'mean_reading_time_seconds']\n",
"\n",
"# Declare result var: type (possible values \"string\", \"number\", \"dataframe\", \"plot\"). Examples: { \"type\": \"string\", \"value\": f\"The highest salary is {highest_salary}.\" } or { \"type\": \"number\", \"value\": 125 } or { \"type\": \"dataframe\", \"value\": pd.DataFrame({...}) } or { \"type\": \"plot\", \"value\": \"temp_chart.png\" }\n",
"result = {\n",
" \"type\": \"dataframe\",\n",
" \"value\": grouped\n",
"}\n",
"```\n",
"In this code, we first convert the `start_local` column to datetime format using `pd.to_datetime()`. Then, we group the data by month on the `start_local` column using `groupby()` and calculate the mean of `reading_time_seconds` for each group. We reset the index of the grouped dataframe using `reset_index()` and rename the columns to `month` and `mean_reading_time_seconds`.\n",
"\n",
"Finally, we declare the `result` variable as a dictionary with type `dataframe` and value as the grouped dataframe.\n",
"The error is raised because the 'start\\_local' column contains null values, and pandas cannot apply the `dt` accessor to non-datetimelike values. We can fill the null values with a default date like the first day of the month before grouping. Here's the corrected code:\n",
"\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs[0]['start_local'] = pd.to_datetime(dfs[0]['start_local'], errors='coerce') # Coerce errors to NaT\n",
"dfs[0]['start_local'] = dfs[0]['start_local'].fillna(dfs[0]['start_local'].min() - pd.Timedelta(days=dfs[0]['start_local'].dt.days.min()))\n",
"dfs[0]['start_local'] = dfs[0]['start_local'].dt.to_period('M') # Convert to period for grouping by month\n",
"grouped = dfs[0].groupby(dfs[0]['start_local']).mean()\n",
"grouped.reset_index(inplace=True)\n",
"grouped.columns = ['month', 'mean_reading_time_seconds']\n",
"\n",
"# Declare result var: type (possible values \"string\", \"number\", \"dataframe\", \"plot\"). Examples: { \"type\": \"string\", \"value\": f\"The highest salary is {highest_salary}.\" } or { \"type\": \"number\", \"value\": 125 } or { \"type\": \"dataframe\", \"value\": pd.DataFrame({...}) } or { \"type\": \"plot\", \"value\": \"temp_chart.png\" }\n",
"result = {\n",
" \"type\": \"dataframe\",\n",
" \"value\": grouped\n",
"}\n",
"```\n",
"\n",
"This code first converts the 'start\\_local' column to datetime format and fills null values with the first day of the month. Then, it converts the 'start\\_local' column to a month period for grouping and calculates the mean 'reading\\_time\\_seconds' for each month.\n",
"The error is being caused by trying to apply the `mean` function to the entire dataframe instead of just the `reading_time_seconds` column. Here is the corrected code:\n",
"\n",
"```python\n",
"# TODO: import the required dependencies\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"dfs[0]['start_local'] = pd.to_datetime(dfs[0]['start_local'])\n",
"grouped = dfs[0].groupby(dfs[0]['start_local'].dt.month)['reading_time_seconds'].mean()\n",
"grouped.reset_index(inplace=True, level=0, drop=True)\n",
"grouped.columns = ['month', 'mean_reading_time_seconds']\n",
"\n",
"# Declare result var: type (possible values \"string\", \"number\", \"dataframe\", \"plot\"). Examples: { \"type\": \"string\", \"value\": f\"The highest salary is {highest_salary}.\" } or { \"type\": \"number\", \"value\": 125 } or { \"type\": \"dataframe\", \"value\": pd.DataFrame({...}) } or { \"type\": \"plot\", \"value\": \"temp_chart.png\" }\n",
"result = {\n",
" \"type\": \"dataframe\",\n",
" \"value\": grouped\n",
"}\n",
"```\n",
"\n",
"This corrected code first converts the `start_local` column to datetime format, then groups the dataframe by the month of the `start_local` column and calculates the mean of the `reading_time_seconds` column. The `reset_index` method is used to reset the index of the resulting dataframe, and the `level` and `drop` parameters are used to remove the multi-index created by the groupby operation. The column names are then updated to match the desired output format.\n"
]
},
{
"data": {
"text/plain": [
"'Unfortunately, I was not able to answer your question, because of the following error:\\n\\nCan only use .dt accessor with datetimelike values\\n'"
]
},
"execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df.chat('group by month on start_local column, using mean on reading_time_seconds')"
]
},
{
"cell_type": "markdown",
"id": "a0ce2049-91cb-4b55-8f65-850de87b5366",
"metadata": {},
"source": [
"# mistral-medium"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "b559033a-1748-40e4-a6ba-20e94c492ee3",
"metadata": {},
"outputs": [],
"source": [
"llm = LlmLLM(api_key=os.getenv(\"MISTRAL_API_KEY\"), model=\"mistral-medium\")"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "1965ee09-d3dc-4954-a2f8-a5293990b0b1",
"metadata": {},
"outputs": [],
"source": [
"df = pd.read_csv(f\"../output/2024-01-19.csv\", parse_dates=[\"day\", \"start_utc\", \"end_utc\"])\n",
"df = SmartDataframe(df, config={\"llm\": llm})"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "9aea45ab-b196-4a50-a7ba-ab9db1f020b5",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Here's the corrected code:\n",
"\n",
"```python\n",
"import pandas as pd\n",
"\n",
"# Convert the 'start\\_local' column to datetime\n",
"dfs[0]['start_local'] = pd.to_datetime(dfs[0]['start_local'])\n",
"\n",
"# Group by month and calculate the mean of 'reading\\_time\\_seconds'\n",
"grouped = dfs[0].groupby(dfs[0]['start_local'].dt.month)['reading_time_seconds'].mean()\n",
"\n",
"# Reset index and rename columns\n",
"grouped.reset_index(inplace=True)\n",
"grouped.columns = ['month', 'mean_reading_time_seconds']\n",
"\n",
"# Define the result variable\n",
"result = {\n",
" \"type\": \"dataframe\",\n",
" \"value\": grouped\n",
"}\n",
"```\n",
"\n",
"The main issue was that you were trying to group by the entire DataFrame instead of just the 'reading\\_time\\_seconds' column. Also, you don't need to group by the entire 'start\\_local' datetime column, but just the month part of it. I fixed these issues and also removed the unnecessary import statement.\n",
"The error is occurring because the `start_local` column contains missing values and the `.dt` accessor cannot be used with non-datetime values. To fix this, you can fill the missing values in the `start_local` column before converting it to datetime. Here's the updated code:\n",
"```\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"# Fill missing values with the first non-missing value\n",
"dfs[0]['start_local'] = dfs[0]['start_local'].fillna(dfs[0]['start_local'].iloc[0])\n",
"\n",
"# Convert start_local to datetime\n",
"dfs[0]['start_local'] = pd.to_datetime(dfs[0]['start_local'])\n",
"\n",
"# Group by month and calculate mean of reading_time_seconds\n",
"grouped = dfs[0].groupby(dfs[0]['start_local'].dt.month).mean()\n",
"\n",
"# Reset index and rename columns\n",
"grouped.reset_index(inplace=True)\n",
"grouped.columns = ['month', 'mean_reading_time_seconds']\n",
"\n",
"# Declare result var: type (possible values \"string\", \"number\", \"dataframe\", \"plot\"). Examples: { \"type\": \"string\", \"value\": f\"The highest salary is {highest_salary}.\" } or { \"type\": \"number\", \"value\": 125 } or { \"type\": \"dataframe\", \"value\": pd.DataFrame({...}) } or { \"type\": \"plot\", \"value\": \"temp_chart.png\" }\n",
"result = {\n",
" \"type\": \"dataframe\",\n",
" \"value\": grouped\n",
"}\n",
"```\n",
"Note that this code fills missing values with the first non-missing value in the `start_local` column. You may want to choose a more appropriate strategy for handling missing values depending on the context of your data.\n"
]
},
{
"data": {
"text/plain": [
"'Unfortunately, I was not able to answer your question, because of the following error:\\n\\nCan only use .dt accessor with datetimelike values\\n'"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df.chat('group by month on start_local column, using mean on reading_time_seconds')"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "e05f6ddb-6991-43d6-b3a8-84c9937b8f27",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"The error occurs because `dfs[0].columns` returns a pandas Index object, not a DataFrame object. To fix the issue, you can modify the code as follows:\n",
"```\n",
"import pandas as pd\n",
"\n",
"# Write code here\n",
"# Split the input string by newline character and filter out empty strings\n",
"data_lines = [line for line in df.split('\\n') if line]\n",
"\n",
"# Create a list of dataframes by reading each line as a csv string\n",
"dfs = [pd.read_csv(pd.compat.StringIO(line), parse_dates=['day', 'start_local', 'start_utc', 'end_local', 'end_utc']) for line in data_lines]\n",
"\n",
"# Extract the columns of the first dataframe and convert them to a list\n",
"columns = list(dfs[0].columns)\n",
"\n",
"result = {\"type\": \"list\", \"value\": columns}\n",
"```\n",
"This code creates a list of dataframes by reading each line in the input string as a csv string. It then extracts the columns of the first dataframe and converts them to a list. Finally, it returns the list of columns in the `result` dictionary.\n",
"Here's the fixed code:\n",
"\n",
"```python\n",
"import pandas as pd\n",
"\n",
"# Assuming that the input data is stored in a string variable called `data`\n",
"data = \"\"\"\n",
"day,start_local,start_utc,end_local,end_utc,reading_time_seconds\n",
"2017-09-16,2017-09-16 18:54:12+02:00,2017-09-16 16:54:12+00:00,2017-09-17 00:21:55+02:00,2017-09-16 22:21:55+00:00,19663.0\n",
"2017-09-17,,,2017-09-17 15:23:55+02:00,2017-09-17 13:23:55+00:00,\n",
"2017-09-15,,,2017-09-16 01:54:41+02:00,2017-09-15 23:54:41+00:00,\n",
"\"\"\"\n",
"\n",
"# Read the CSV data into a pandas DataFrame\n",
"dfs = [pd.read_csv(pd.compat.StringIO(datum), parse_dates=['day', 'start_local', 'start_utc', 'end_local', 'end_utc']) for datum in data.strip().split('\\n') if datum]\n",
"\n",
"# Extract the column names from the first DataFrame in the list\n",
"result = {\"type\": \"dataframe\", \"value\": dfs[0].columns.tolist()}\n",
"```\n",
"\n",
"Note that I have assumed that the input data is stored in a string variable called `data`. If the data is coming from a different source, you will need to modify the code accordingly.\n",
"\n",
"Also, note that I have used the `strip()` method to remove any leading or trailing whitespace from the input string, and then split it into lines using the `\\n` newline character. This is necessary to correctly split the string into separate CSV records.\n",
"\n",
"Finally, I have used the `tolist()` method to convert the `Index` object returned by `dfs[0].columns` to a list of column names. This is necessary because the `value` field of the `result` dictionary is expected to be a list.\n",
"\n",
"I hope this helps! Let me know if you have any further questions.\n"
]
},
{
"data": {
"text/plain": [
"\"Unfortunately, I was not able to answer your question, because of the following error:\\n\\n'list' object has no attribute 'to_json'\\n\""
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df.chat('show the columns')"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "16c883cd-49b9-4dd9-b34d-a0b97811be30",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.5"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment