Decision Weights in Prospect Theory

Overview

Decision Weights in Prospect Theory

It's clear that humans are irrational, but how irrational are they? After some research into behavourial economics, I became very interested in Prospect Theory (see Chapter 29 of Thinking, Fast and Slow). A very interesting part of Prospect theory is that it is not probabilities that are used in the calculation of expected value:

ev

Here, the q's are not the probabilities of outcome z, but it is from another probability measure called decision weights that humans actually use to weigh outcomes. Using a change of measure, we can observe the relationship between the actual probabilities and the decision weights:

cmg

My interest is in this change of measure.

The Setup

Suppose you have two choices:

  1. Lottery A: have a 1% chance to win $10 000,
  2. Lottery B: have a 99% chance to win $101

Which would you prefer?

Well, under the real world probabilty measure, these two choices are equal: .99 x 101 = .01 x 10000. Thus a rational agent would be indifferent to either option. But a human would have a preference: they would see one more valuable than the other. Thus:

inq

rewritten:

inq2

and dividing:

inq3

What's left to do is determine the direction of the first inequality.

Mechanical Turk it.

So I created combinations of probabilities and prizes, all with equal real-world expected value, and asked Turkers to pick which one they preferred. Example:

Imgur

Again, notice that .5 x $200 = .8 x $125 = $100. The original HIT data and the python scripts that generate are in the repo, plus the MTurk data. Each HIT received 10 turkers.

Note: The Turking cost me $88.40, if you'd like to give back over Gittip, that would be great =)

Note: I called the first choice Lottery A and the second choice Lottery B.

Analysis

Below is a slightly inappropriate heatmap of the choices people made. If everyone was rational, and hence indifferent to the two choices, the probabilities should hover around 0.5. This is clearly not the case.

Imgur

What else do we see here?

  1. As expected, people are loss averse: every point in the lower-diagonal is where lottery A had a high probability of success than B. The matrix shows that most points in here are greater than 50%, thus people chose the safer bet more often.
  2. The exception to the above point is the fact that 1% is choosen more favourably over 2%. This is an instance of the possibility effect. People are indifferent between 1% and 2%, as they are both so rare, thus will pick the one with larger payoff.

FAQ

  1. Why did I ask the Turkers to deeply imagine winning $50 dollars before answering the question? This was to offset a potential anchoring effect: if a Turkers first choice had prize $10 000, then any other prize would have looked pitiful, as the anchor had been set at $10 000. By having them imagine winning $50 (lower than any prize), then any prize they latter saw would appear better than this anchor.

  2. Next steps? I'd like to try this again, with more control over the Turkers (have a more diverse set of Turkers on it).

This data is mirrored and can be queried via API here

Owner
Cameron Davidson-Pilon
CEO of Pioreactor. Former Director of Data Science @Shopify. Author of Bayesian Methods for Hackers and DataOrigami.
Cameron Davidson-Pilon
A simple machine learning python sign language detection project.

SST Coursework 2022 About the app A python application that utilises the tensorflow object detection algorithm to achieve automatic detection of ameri

Xavier Koh 2 Jun 30, 2022
An easier way to build neural search on the cloud

Jina is geared towards building search systems for any kind of data, including text, images, audio, video and many more. With the modular design & multi-layer abstraction, you can leverage the effici

Jina AI 17k Jan 01, 2023
flexible time-series processing & feature extraction

A corona statistics and information telegram bot.

PreDiCT.IDLab 206 Dec 28, 2022
Code Repository for Machine Learning with PyTorch and Scikit-Learn

Code Repository for Machine Learning with PyTorch and Scikit-Learn

Sebastian Raschka 1.4k Jan 03, 2023
A concept I came up which ditches the idea of "layers" in a neural network.

Dynet A concept I came up which ditches the idea of "layers" in a neural network. Install Copy Dynet.py to your project. Run the example Install matpl

Anik Patel 4 Dec 05, 2021
Causal Inference and Machine Learning in Practice with EconML and CausalML: Industrial Use Cases at Microsoft, TripAdvisor, Uber

Causal Inference and Machine Learning in Practice with EconML and CausalML: Industrial Use Cases at Microsoft, TripAdvisor, Uber

EconML/CausalML KDD 2021 Tutorial 124 Dec 28, 2022
Machine Learning Algorithms ( Desion Tree, XG Boost, Random Forest )

implementation of machine learning Algorithms such as decision tree and random forest and xgboost on darasets then compare results for each and implement ant colony and genetic algorithms on tsp map,

Mohamadreza Rezaei 1 Jan 19, 2022
Covid-polygraph - a set of Machine Learning-driven fact-checking tools

Covid-polygraph, a set of Machine Learning-driven fact-checking tools that aim to address the issue of misleading information related to COVID-19.

1 Apr 22, 2022
Repositório para o #alurachallengedatascience1

1° Challenge de Dados - Alura A Alura Voz é uma empresa de telecomunicação que nos contratou para atuar como cientistas de dados na equipe de vendas.

Sthe Monica 16 Nov 10, 2022
Python package for machine learning for healthcare using a OMOP common data model

This library was developed in order to facilitate rapid prototyping in Python of predictive machine-learning models using longitudinal medical data from an OMOP CDM-standard database.

Sontag Lab 75 Jan 03, 2023
Winning solution for the Galaxy Challenge on Kaggle

Winning solution for the Galaxy Challenge on Kaggle

Sander Dieleman 483 Jan 02, 2023
Little Ball of Fur - A graph sampling extension library for NetworKit and NetworkX (CIKM 2020)

Little Ball of Fur is a graph sampling extension library for Python. Please look at the Documentation, relevant Paper, Promo video and External Resour

Benedek Rozemberczki 619 Dec 14, 2022
inding a method to objectively quantify skill versus chance in games, using reinforcement learning

Skill-vs-chance-games-analysis - Finding a method to objectively quantify skill versus chance in games, using reinforcement learning

Marcus Chiam 4 Nov 19, 2022
A Python implementation of GRAIL, a generic framework to learn compact time series representations.

GRAIL A Python implementation of GRAIL, a generic framework to learn compact time series representations. Requirements Python 3.6+ numpy scipy tslearn

3 Nov 24, 2021
Tangram makes it easy for programmers to train, deploy, and monitor machine learning models.

Tangram Website | Discord Tangram makes it easy for programmers to train, deploy, and monitor machine learning models. Run tangram train to train a mo

Tangram 1.4k Jan 05, 2023
Esse é o meu primeiro repo tratando de fim a fim, uma pipeline de dados abertos do governo brasileiro relacionado a compras de contrato e cronogramas anuais com spark, em pyspark e SQL!

Olá! Esse é o meu primeiro repo tratando de fim a fim, uma pipeline de dados abertos do governo brasileiro relacionado a compras de contrato e cronogr

Henrique de Paula 10 Apr 04, 2022
Automated Machine Learning Pipeline with Feature Engineering and Hyper-Parameters Tuning

The mljar-supervised is an Automated Machine Learning Python package that works with tabular data. I

MLJAR 2.4k Jan 02, 2023
Tools for mathematical optimization region

Tools for mathematical optimization region

林景 15 Nov 30, 2022
A Python Module That Uses ANN To Predict A Stocks Price And Also Provides Accurate Technical Analysis With Many High Potential Implementations!

Stox A Module to predict the "close price" for the next day and give "technical analysis". It uses a Neural Network and the LSTM algorithm to predict

Stox 31 Dec 16, 2022