Yazi Repo 2.0

Explore

24 April

Lwanda's big day

Website

It looks like there are still some full width graphs. Have you made them all small? It maybe because there is no text next to them to reduce though

Coding it up

I want you to create python code that creates 2 tables, one is for straight-line checking/analysis and the other is the answers with fail columns

You need to convert raw survey responses into a table with user IDs going down the rows and question heading columns.

This task is all about data accuracy - we are checking timing and answer pattern. Our 2 resources

Survey answers URL:

https://bot-yazi.comparisure.co.za/api/survey/109/report/?from=2023-04-11T00:00&to=2023-04-30T00:00⁠

⁠

JSON code

⁠

109 JSON code.json

18.5 kB

⁠

JSON Code extract example

Here is a question example to show you how it is structured:

How has loadshedding affected your ice-cream promotions, i.e. Sell ice-cream tubs in bulk or on promotions

(i.e. 2 for the price of 1 deals)?

Much less open to ice-cream promotions

Slightly less open to ice-cream promotions

No impact - same as before

Slightly more open to ice-cream promotions

Much more open to ice-cream promotions

Tables to create

Straight lining - Create a new table that converts raw survey responses into a table with user IDs going down the rows and question heading columns. But, all the multiple choice answers ("type": "multiple",) need to be numbers - the number the answer was in the question. I.e. if someone said: “No impact - same as before” in above then the answer in the column for that question would be 3. The way you will code this is by looking up the responseQuestionId in the raw response data URL against the sequence in the JSON code file shared above.

Make userId in column A

Questions must be in ascending order (see my prompt in chatgpt - link below)

I am unsure exactly how to assess this so the best thing might be to ask chatgpt the best method to detect fraudsters or people who are negligent and answer the same

Answers table - Create a new table that converts raw survey responses into a table with user IDs going down the rows and question heading column. Then add the following columns

Columns to add to Answers table:

Average response time (sec) - You must calculate the average time between each response per user in seconds

Response time check - If response ave time is less than 7sec then mark as Fail, otherwise Pass

Straight lines - count how many times the answer sequence is the same for 3 answers. E.g the following would output 2 as A’s and E’s appear 3 or more times in answers: A, A, A, B, C,D, E, E, E, E, D

Straight-line checker - if there are 2 or more repeats of 3 then Fail, otherwise Pass

Straight-lining table

Straight-lining table

User ID

question 1

Question 2

Question 3

283747

87474

There are no rows in this table

⁠

Example table

Example table

User ID

question 1

Question 2 (etc.)

Question 3 (etc.)

Average response time (sec)

Response time check

Straight lines

283747

spaza

red

Hard

Pass

87474

supermarket

blue

Easy

Fail

There are no rows in this table

⁠

Prompts I have already give that may help

⁠

https://chat.openai.com/c/a327714d-c472-4590-83a3-ad1a4b628896⁠

⁠

⁠

Great, now I want you to turn all multiple choice options into numbers. The number the answer will be turned into will be based on the order that the option was presented to the respondent in.

Below options is where the multiple choice options are displayed. They are displayed in the order that is presented to the survey respondent. Here is an example to show

How has loadshedding affected your ice-cream promotions, i.e. Sell ice-cream tubs in bulk or on promotions

(i.e. 2 for the price of 1 deals)?

Much less open to ice-cream promotions

Slightly less open to ice-cream promotions

No impact - same as before

Slightly more open to ice-cream promotions

Much more open to ice-cream promotions

Prompt idea for straight lining

Convert responseAnswer into the placement number or order number that it is listed in, in the actual survey. To know what number to convert it to, we need to look it up using the 109 JSON code.json file and lookup the answer using the responseQuestionId in the survey response table with the questionKey in the JSON

Old Prompt

I made a mistake, I want you to arrange it in python, not sheets (going back to our previous step in replit)

The data is in the following format:

- Column A: userId

- Column B: userName

- Column C: responseTimestamp

- Column D: responseQuestionId

- Column E: responseAnswer

I want you to reformat the data so that each question ID is a heading in each column and each row is a unique user

I also want to add 2 time stamp columns:

- Column B: Survey start

- Column C: Survey end

Survey start value: the earliest time stamp (from responseTimestamp) from a user

Survey end value: the last time stamp (from responseTimestamp) from a user

old prompt 2

Great. Now I want us to update the code so that the columns are in a specific order using responseQuestionId as the attribute to sort on.

responseQuestionId is contains a string with a number in the middle of the string after the letter 'q'. E.g. '109_q10_mcq' & '109_q13_mcq'. In this instance, the numbers are 10 & 13

I want you so sort the columns in ascending order based on the number after the letter q. Add this to the python code above

AutoGPT

Create agents in first 2 videos (replit and nick renotte

Tools needed (well, I am not 100% sure so you can also figure out but seems like)

Visual studio

Replit

AutoGPT videos

AutoGPT videos

Name

URL

Length

Notes

Status

LangChain Crash Course: Build a AutoGPT app in 25 minutes!

27.27

great vid - seems like a good Tuber

I want you to implement exactly what he does

What can you do with Autonomous Agents? Auto GPT & Replit = The fastest way to spin up an Agent

replit → seems insanely quick to do

I want you to try too

AutoGPT Tutorial - Create Your Own AI Agents!

12.25

Set limits on usage

Create agent

Not sure what IDE

Agent GPT deploys auto gpt on the browser | AgentGPT - Beginners tutorial | No Code AI agents

12.3

uses Agent GPT (pretty simple looks like

He asks to build AI newsletter

AutoGPT Tutorial - More Exciting Than ChatGPT

23.55

on mac which might not be ideal

There are no rows in this table

⁠

OpenAI: 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

⁠

https://platform.openai.com/⁠

⁠

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.