📈 SiRE (Simulation-Informed Revenue Extrapolation with Confidence Estimate for Scaleup Companies Using Scarce Time-Series Data), accepted by CIKM'2022 🗽

Overview

Simulation-informed Revenue Extrapolation (SiRE) with confidence estimate for scaleup companies using scarce timeseries data.


OverviewUsageExampleArxivCitation

version cikm python python

Overview

Investment professionals rely on extrapolating company revenue into the future (i.e. revenue forecast) to approximate the valuation of scaleups (private companies in a high-growth stage) and inform their investment decision. This task is manual and empirical, leaving the forecast quality heavily dependent on the investment professionals’ experiences and insights. Furthermore, financial data on scaleups is typically proprietary, costly and scarce, ruling out the wide adoption of data-driven approaches.

We propose an algorithm SiRE (Simulation-informed Revenue Extrapolation) that generates fine-grained long-term revenue predictions on small datasets and short time-series. SiRE fulfills eight important practical requirements:

  1. it performs well in multiple business sectors;
  2. it works on small datasets with only a few hundred scaleups;
  3. the extrapolation can commence from short revenue time-series;
  4. it should produce a fine-grained time-series of at least three-year length;
  5. each predicted revenue point should come with a confidence estimation;
  6. it does not require any alternative data other than sector information;
  7. the model can be timely and effortlessly adapted to any data change.
  8. the prediction is explainable.

Usage

The source code and two datasets (transformed and anonymized) are already bundled in this repo. To run SiRE on both datasets, follow the steps below.

  • To isolate the Python runtime for SiRE experimentation, install a virtual environment such as Anaconda. We hereby assume that you are using Anaconda. The following steps should be run in the project folder sire from a Terminal.

  • Create a virtual Python environment by running:

$ conda create --name sire python=3.8
$ conda activate sire
  • Install the dependent libraries:
$ pip install -r requirements.txt
  • Start Python Notebook:
$ jupyter notebook

To experiment SiRE on SapiQ dataset, run through the notebook named sire_sapiq.ipynb.

To experiment SiRE on ARR129 dataset, run through the notebook named sire_arr129.ipynb.

Example Forecast

Following is the forecast (from 2018-05-01) visualization for company in ARR129 with ID 5cf74ff84639c79510ef8491d4725967336ad3627216a50553f030f9c51d93bc.

drawing

The corresponding predicted revenue points with 95% confidence interval (95CI) look like below.

DATE MEAN 95CI_low 95CI_high MIN MAX
2018-05-01 52.6052 51.6823 53.5282 50.6054 54.8298
2018-06-01 55.3628 54.0489 56.6767 52.6138 58.5401
2018-07-01 57.8639 56.1423 59.5856 54.3638 62.0497
2018-08-01 60.2249 58.0926 62.3571 55.9818 65.4445
2018-09-01 62.5506 60.0036 65.0977 57.5778 68.819
2018-10-01 64.9284 61.9523 67.9044 59.2417 72.2662
... ... ... ... ... ...
2021-02-01 298.059 254.498 341.621 240.334 416.363
2021-03-01 314.914 268.262 361.567 253.592 442.045
2021-04-01 331.436 281.516 381.356 266.558 468.107

Cite This Work

@inproceedings{cao-etal-2022-sire,
    title = "Simulation-Informed Revenue Extrapolation with Confidence Estimate for Scaleup Companies Using Scarce Time Series Data",
    author = "Cao, Lele  and
      Horn, Sonja  and
      von Ehrenheim, Vilhelm  and
      Stahl, Richard Anselmo  and
      Landgren, Henrik",
    booktitle = "Proceedings of the 31st ACM International Conference on Information and Knowledge Management (CIKM ’22), October 17–21, 2022, Atlanta, GA, USA",
    month = oct,
    year = "2022",
    address = "New York, NY, USA",
    publisher = "Association for Computing Machinery (ACM)",
    pages = "12 pages",
    doi = "10.1145/3511808.3557110"
}
You might also like...

This repo using OpenVINO POT tool to quantize yolo-v7 series or yolo-v5 series onnx model to IR model

Yolo v7 / Yolo v5 Quantization using OpenVINO POT This repo using OpenVINO POT tool to quantize yolo-v7 series or yolo-v5 series onnx model to IR mode

Nov 7, 2022

Data Cleaning. Data Integration. Data Reduction for 1-Data Quality. 2-Data Transformation. 3-Data Mining. 4-Pattern Evaluation. 5-Representing Knowledge in Data Mining.

Data Cleaning. Data Integration. Data Reduction for 1-Data Quality. 2-Data Transformation. 3-Data Mining. 4-Pattern Evaluation. 5-Representing Knowledge in Data Mining.

Python_application_for_dataMining Oreview: We aim to achieve a prediction model for improving data analysis and reporting. The programming language th

Sep 21, 2022

Bayesian Inference of Slide-level Confidence via Uncertainty Thresholding

Bayesian Inference of Slide-level Confidence via Uncertainty Thresholding

BISCUIT Bayesian Inference of Slide-level Confidence via Uncertainty Index Thresholding Requirements Tensorflow =2.7 (and associated pre-requisites)

Nov 22, 2022

Anytime Dense Prediction with Confidence Adaptivity (ICLR 2022)

Anytime Dense Prediction with Confidence Adaptivity (ICLR 2022)

Anytime Dense Prediction with Confidence Adaptivity Official PyTorch implementation for the following paper: Anytime Dense Prediction with Confidence

Nov 22, 2022

[ECCV 2022] Contrastive Prototypical Network with Wasserstein Confidence Penalty

Contrastive Prototypical Network with Wasserstein Confidence Penalty PyTorch implementation of Contrastive Prototypical Network with Wasserstein Confi

Sep 8, 2022

PyTorch code for DeepTIMe: Deep Time-Index Meta-Learning for Non-Stationary Time-Series Forecasting

DeepTIMe: Deep Time-Index Meta-Learning for Non-Stationary Time-Series Forecasting Figure 1. Overall approach of DeepTIMe. Official PyTorch code repos

Dec 29, 2022

A web application that would provide a public, accessible resource that helps educate consumers about the brands they support so they can make well-informed, more ethical decisions.

A web application that would provide a public, accessible resource that helps educate consumers about the brands they support so they can make well-informed, more ethical decisions.

hackKU3.0 Hackathon Project for HackKU 2022 Clarity Inspiration Our ultimate vision is to create a platform where users can look up a company's name a

Aug 24, 2022

Code for the paper "Thermodynamics-informed graph neural networks" published in IEEE Transactions on Artificial Intelligence (TAI).

Code for the paper

Thermodynamics-informed graph neural networks Abstract In this paper we present a deep learning method to predict the temporal evolution of dissipativ

Dec 4, 2022

Physics informed neural network (PINN) for the 1D Transport equation with decay

Transport Equation with decay 1D Physics informed neural network (PINN) for the 1D Transport equation with decay This module implements the Physics In

May 31, 2022
Releases(0.0.1)
Owner
EQT
EQT
The paper "Learning Perceptual Position-aware Shapelet for Time Series" accepted by ECML PKDD 2022

Code The paper "Learning Perceptual Position-aware Shapelet for Time Series" accepted by ECML PKDD 2022 Fig. 1: General Architecture of Perceptual Pos

Minh-Tuan Tran 1 Jun 19, 2022
Companies house base data and "persons with signficant control" to FollowTheMoney converter

UK Companies House PSC Parser Convert the UK companies house base data product and PSC (persons with significant control) to FollowTheMoney format. Th

OpenSanctions.org 2 Dec 15, 2022
A data structure to track data over time. It works by tracking time/schedule information rather than tracking data changes over time.

pcontract A data structure to track data over time. It works by tracking time/schedule information rather than tracking data changes over time. Reason

şuayip üzülmez 9 Oct 22, 2022
🎓🖥️ Solutions for 350+ Interview Questions asked at FAANG and other top tech companies

Daily Coding Problem Solutions This repository contains the Daily Coding Problem solutions in python. Setup Install the 3rd party packages, perform th

Happy Life 4 Oct 3, 2022
Italia Opensource is a list of open source projects created by Italian companies or developers. The repository intends to give visibility to open source projects and stimulate the community to contribute to growing the ecosystem.

Italia Opensource Italia Opensource is a list of open source projects created by Italian companies or developers. The repository intends to give visib

Italia Opensource 143 Jan 3, 2023
Automation in Python, for billing and sales calculation for small and medium-sized companies, used for managing goals and commissions.

Automação de gerenciamento do faturamento em Python ?? ?? Project created in carrying out a skill test with web development Portugues · English Financ

Gilles 2 Nov 24, 2022
PyTorch implementation of Denoising of 3D MR images using a voxel-wise hybrid residual MLP-CNN model to improve small lesion diagnostic confidence (MICCAI 2022).

Denoising of 3D MR images using a voxel-wise hybrid residual MLP-CNN model to improve small lesion diagnostic confidence PyTorch implementation of Den

null 9 Nov 15, 2022
Physics Informed Neural Fields for Smoke Reconstruction with Sparse Data

The pytorch implementation of the paper, Physics Informed Neural Fields for Smoke Reconstruction with Sparse Data (SIGGRAPH 2022 https://people.mpi-inf.mpg.de/~mchu/projects/PI-NeRF/), from M. Chu, L. Liu, Q. Zheng, E. Franz, H.P. Seidel, C. Theobalt, R. Zayer.

Mengyu (Rachel) Chu 15 Dec 19, 2022
NannyML is an open-source python library that allows you to estimate post-deployment model performance

NannyML is an open-source python library that allows you to estimate post-deployment model performance (without access to targets), detect data drift, and intelligently link data drift alerts back to changes in model performance. Built for data scientists, NannyML has an easy-to-use interface, interactive visualizations, is completely model-agnostic and currently supports all tabular binary classification use cases.

null 1.3k Jan 1, 2023
For Students taking AI as a college course, the Romania Roadmap problem is a common example used to implement Informed Heuristic Searches. This repository has solutions for this problem using a few of the said search strategies

Romania Road Map Problem For Students taking AI as a college course, the Romania Roadmap problem is a common example used to implement Informed Heuris

null 1 Sep 20, 2022