Contact
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutSign UpSign In
| Download
Views: 100
Image: ubuntu2204-dev
Kernel: Python 3 (system-wide)

OpenAI API in CoCalc

  1. Sign up to OpenAI and get your API key.

  2. Open CoCalc's Project Settings and scroll to the "Custom environment variables" section. Add your key there like:

    {
  "OPENAI_API_KEY": "sk-1942.......................AWEf"
}

  3. Restart your project to make that environment variable be set.

Load some packages, openapi is already installed …

import os
from IPython.display import Markdown
import openai # already installed
import pkg_resources
pkg_resources.get_distribution('openai').version
'0.27.2'

Load your API key from that environment variable

openai.api_key = os.getenv("OPENAI_API_KEY")

Quick test, completing a prompt

prompt = """\
Tell me 5 cool things I can do with the programming language Python:
"""

response = openai.Completion.create(model="text-davinci-003",
                                    prompt=prompt,
                                    temperature=.7,
                                    max_tokens=100)

print(response["choices"][0]["text"])
1. Build web applications with frameworks like Django and Flask. 2. Create powerful data analysis applications with libraries like Pandas and NumPy. 3. Develop GUI programs for desktop and mobile devices using Tkinter and PyQt. 4. Create games with the Pygame library. 5. Use Python to automate tedious tasks with its powerful scripting capabilities.

Chatting with GPT-4

As of today, you need to have access to this model…

answer = openai.ChatCompletion.create(
    model="gpt-4",
    messages=[{
        "role":
        "system",
        "content":
        "Formulate your answer to a student at a university, who knows a bit of Python."
    }, {
        "role":
        "user",
        "content":
        "Show me how to plot this sequence of numbers in Python and extrapolate it a little bit using statsmodels: 2, 3, 15, 14, 9.2, 11, 9, 12.1, 15, 16"
    }])

answer["choices"]
[<OpenAIObject at 0x7f0aa4d3e250> JSON: { "finish_reason": "stop", "index": 0, "message": { "content": "To plot the given sequence of numbers and extrapolate it using the statsmodels library, you can use the following steps:\n\n1. Install the required libraries: statsmodels and matplotlib (if you haven't done so)\n2. Import the required libraries and functions.\n3. Load the data.\n4. Fit the model using statsmodels' SARIMAX function.\n5. Predict the extrapolated values.\n6. Plot the results.\n\nHere's the code:\n\n```python\n# Step 1: Install libraries (run this line in your command line, not in the Python file)\n# pip install statsmodels matplotlib\n\n# Step 2: Import required libraries and functions\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport statsmodels.api as sm\n\n# Step 3: Load the data\ndata = np.array([2, 3, 15, 14, 9.2, 11, 9, 12.1, 15, 16])\n\n# Step 4: Fit the model (SARIMAX)\n# Order and seasonal_order parameters are (p, d, q) and (P, D, Q, s) respectively, chosen by trial and error\n# Here, we use an ARIMA model with order (1, 0, 0), which means 1 autoregressive, 0 differencing and 0 moving average component\nmodel = sm.tsa.SARIMAX(data, order=(1, 0, 0), seasonal_order=(0, 0, 0, 0))\nresults = model.fit()\n\n# Step 5: Predict the extrapolated values (here, we extrapolate 5 new values)\nextrapolate_steps = 5\npredictions = results.get_prediction(start=len(data), end=len(data) + extrapolate_steps - 1, dynamic=True)\n\n# Step 6: Plot the results\nplt.plot(data, label='Original data')\nplt.plot(np.arange(len(data), len(data) + extrapolate_steps), predictions.predicted_mean, label='Extrapolated data')\nplt.xlabel('Time step')\nplt.ylabel('Value')\nplt.legend()\nplt.show()\n```\n\nThis code will generate a plot showing the original data and the extrapolated data using an ARIMA model from the statsmodels library. Note that you might need to adjust the ARIMA model's order and seasonal_order parameters depending on your specific time series.", "role": "assistant" } }]

Neatly display the answer using IPython's Markdown object

Markdown(answer["choices"][0]['message']["content"])

To plot the given sequence of numbers and extrapolate it using the statsmodels library, you can use the following steps:

  1. Install the required libraries: statsmodels and matplotlib (if you haven't done so)

  2. Import the required libraries and functions.

  3. Load the data.

  4. Fit the model using statsmodels' SARIMAX function.

  5. Predict the extrapolated values.

  6. Plot the results.

Here's the code:

# Step 1: Install libraries (run this line in your command line, not in the Python file)
# pip install statsmodels matplotlib

# Step 2: Import required libraries and functions
import numpy as np
import matplotlib.pyplot as plt
import statsmodels.api as sm

# Step 3: Load the data
data = np.array([2, 3, 15, 14, 9.2, 11, 9, 12.1, 15, 16])

# Step 4: Fit the model (SARIMAX)
# Order and seasonal_order parameters are (p, d, q) and (P, D, Q, s) respectively, chosen by trial and error
# Here, we use an ARIMA model with order (1, 0, 0), which means 1 autoregressive, 0 differencing and 0 moving average component
model = sm.tsa.SARIMAX(data, order=(1, 0, 0), seasonal_order=(0, 0, 0, 0))
results = model.fit()

# Step 5: Predict the extrapolated values (here, we extrapolate 5 new values)
extrapolate_steps = 5
predictions = results.get_prediction(start=len(data), end=len(data) + extrapolate_steps - 1, dynamic=True)

# Step 6: Plot the results
plt.plot(data, label='Original data')
plt.plot(np.arange(len(data), len(data) + extrapolate_steps), predictions.predicted_mean, label='Extrapolated data')
plt.xlabel('Time step')
plt.ylabel('Value')
plt.legend()
plt.show()

This code will generate a plot showing the original data and the extrapolated data using an ARIMA model from the statsmodels library. Note that you might need to adjust the ARIMA model's order and seasonal_order parameters depending on your specific time series.

… and now copy/paste/adapt the provided code snippet


# Step 2: Import required libraries and functions
import numpy as np
import matplotlib.pyplot as plt
import statsmodels.api as sm

# Step 3: Load the data
data = np.array([2, 3, 15, 14, 9.2, 11, 9, 12.1, 15, 16])

# Step 4: Fit the model (SARIMAX)
# Order and seasonal_order parameters are (p, d, q) and (P, D, Q, s) respectively, chosen by trial and error
# Here, we use an ARIMA model with order (1, 0, 0), which means 1 autoregressive, 0 differencing and 0 moving average component
model = sm.tsa.SARIMAX(data, order=(1, 0, 0), seasonal_order=(0, 0, 0, 0))
results = model.fit()

# Step 5: Predict the extrapolated values (here, we extrapolate 5 new values)
extrapolate_steps = 5
predictions = results.get_prediction(start=len(data), end=len(data) + extrapolate_steps - 1, dynamic=True)

# Step 6: Plot the results
plt.plot(data, label='Original data')
plt.plot(np.arange(len(data), len(data) + extrapolate_steps), predictions.predicted_mean, label='Extrapolated data')
plt.xlabel('Time step')
plt.ylabel('Value')
plt.legend()
plt.show()
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/statespace/sarimax.py:966: UserWarning: Non-stationary starting autoregressive parameters found. Using zeros as starting parameters. warn('Non-stationary starting autoregressive parameters' This problem is unconstrained.
RUNNING THE L-BFGS-B CODE * * * Machine precision = 2.220D-16 N = 2 M = 10 At X0 0 variables are exactly at the bounds At iterate 0 f= 5.57356D+00 |proj g|= 5.18269D+00 At iterate 5 f= 3.03758D+00 |proj g|= 3.61441D-02 At iterate 10 f= 3.02871D+00 |proj g|= 3.68434D-06 * * * Tit = total number of iterations Tnf = total number of function evaluations Tnint = total number of segments explored during Cauchy searches Skip = number of BFGS updates skipped Nact = number of active bounds at final generalized Cauchy point Projg = norm of the final projected gradient F = final function value * * * N Tit Tnf Tnint Skip Nact Projg F 2 10 11 1 0 0 3.684D-06 3.029D+00 F = 3.0287137752620881 CONVERGENCE: NORM_OF_PROJECTED_GRADIENT_<=_PGTOL
Image in a Jupyter notebook