Modified version of question C ideas
Title C Guaranteed Prize
Data preprocessing
3σ principle
Interval judgment method, artificial judgment
Question 1
Cluster analysis for simple classification
interrelationship
The data obeys the normal distribution (after the distribution type is determined by the KS test, etc.) before the person correlation can be done
Chart combination (heat map, data result table)
Distribution
Macro time law, micro time law, holiday time law,
Question 2
- Determine the commodity demand curve
Linear y=kx+b
Nonlinear Y=a*e^x+b
Y=a*sin(b*x)+c
2. Total daily replenishment volume and pricing strategy
K*Total Sales+b=Cost Plus Pricing=(1+Markup Rate)*Cost Purchase Price
Sales volume, mark-up rate, cost purchase price
- Prediction method Predict an unknown quantity
- Set an unknown quantity (the decision variable of the optimization model)
- Equality relationship to solve the third unknown quantity
Forecasting method (time series, gray forecast)
Question 3
dual-objective optimization
- Maximize the benefits
- meet needs
algorithm
Commonly used algorithms for solving optimization models include the following main types:
Nonlinear Programming (NLP):
Gradient Descent: Used to find local optimal solutions.
Newton's Method: Faster convergence, but requires second derivative information.
Quasi-Newton Methods (Quasi-Newton Methods): Use limited second-order information to approximate the Newton method, common ones include BFGS and DFP algorithms.
Integer Nonlinear Programming (Mixed-Integer Nonlinear Programming, MINLP):
Branch and Bound with NLP: Combining branching for integer problems with nonlinear programming methods.
Dynamic Programming: used to solve optimization problems with recursive structures, usually used for sequence decision-making problems and optimal control problems.
Simulated Annealing: A random search algorithm usually used to solve complex combinatorial optimization problems.
Genetic Algorithms: An optimization algorithm that imitates natural selection and genetic mechanisms and is usually used to solve complex combinatorial optimization problems.
Particle Swarm Optimization: A heuristic optimization algorithm that simulates the group behavior of a flock of birds or a school of fish to solve optimization problems.
Model Predictive Control (MPC): A method for dynamic system control and optimization problems that typically uses mathematical models and optimization to predict future states and make optimal decisions.
Mixed-Integer Linear Programming (MILP) of linear programming: combines the characteristics of linear programming and integer programming and is widely used in combinatorial optimization and planning problems.
Question 4
1. Competitor data: Collect competitors’ sales data and pricing strategies to understand market competition. This helps supermarkets develop more competitive pricing strategies and promotions.
2. Consumer behavior data: Collect consumer shopping behavior data, including purchase history, shopping basket combination, shopping frequency and preferences. This can help supermarkets better understand consumer needs and optimize product mix and promotion strategies.
3. Supply chain data: Understand supply chain information, including supplier delivery times, inventory levels, supplier reliability, etc. This can help supermarkets optimize inventory management and replenishment plans.
4. Market trend data: Collect market trend data, including price trends, seasonal changes, new product launches, etc. This helps supermarkets adjust pricing strategies and sales plans to respond to market changes.
5. Promotional effectiveness data: Track the effectiveness of promotional activities, including discount rates, promotional periods, promotional channels, etc. This can help supermarkets evaluate the effectiveness of promotional activities and optimize promotional strategies.
6. Weather data: Know the weather conditions, especially the weather factors related to vegetable sales, such as temperature, precipitation, etc. Weather has an important impact on vegetable sales, so weather data can help supermarkets better predict sales and inventory needs.