Dynamic pricing is used to create different prices for different customers, based on their location or other circumstances. This method was first used for American Airlines in the year 1980. They figured out that not all customers are the same, they all have different concerns and different priorities. While some customers want tickets at a cheaper price while others want good service. This method is now used in all movie booking systems. There are some parameters, depends on which the price varies, like whether the customer wants to book a ticket in the front row or the back, or the customer wants an executive seat or a couple of seats. Not only the seats, but prices also depend on where the movie hall is located and which show timing the customer has chosen.
All ticket booking apps whether it is PVR or BookMyShow uses dynamic pricing. At times of high demand, when a new movie launches BookMyShow increases the price of the tickets.
The other way is to come up with discounts or to give users with customized offers for a limited period. In other words, dynamic pricing is a prediction problem, where machine learning is the best tool to tackle it.
3. Introduction
Most of the dynamic pricing application area’s problems
share three main characteristics. First, the selling season
lasts for a finite span of time/ period. The products have a
limited and fixed selling season time and they are time-
sensitive. Like in case of airlines industry, reservation of
seats is permitted before some months before the actual
departure time, and in case of fashion industry the selling
and buying of apparels some time only last for six to eight
weeks and similarly movie-tickets are available online for
pre-booking which lasts till the show time.
4. Dynamic Pricing
• Dynamic pricing is used to create different prices for different
customers, based on their location or other circumstances.
• This method was first used for American Airlines in the year 1980.
• They figured out that not all customers are the same, they all have
different concerns and different priorities.
5. Dynamic Pricing
• While some customers want tickets at a cheaper price while others
want good service.
• This method is now used in all movie booking systems.
• There are some parameters, depends on which the price varies, like
whether the customer wants to book a ticket in the front row or the
back, or the customer wants an executive seat or a couple of seats.
• Not only the seats, but prices also depend on where the movie hall is
located and which show timing the customer has chosen
6. Algorithms used for Pricing
1. Linear Regression
2. Gradient Boosting Regression
3. Random Forest Regression
4. Neural Network as Regressive Model
7. Cinema
Comparison
Linear Regression
Training a Linear Regression model with our data gave Rmse (Root mean
square error) of: 22.74. This shows that our model will predict output prices
for show tickets with a mean error of around 22 rupees. This result is not very
satisfying for movies with very low demand as a negative error would lead to
a very low price.
8. Gradient Boosting Regression
* Gradient boosting algorithm is one of the most powerful
algorithms in the field of machine learning.
*As we know that the errors in machine learning algorithms are
broadly classified into two categories i.e. Bias Error and Variance
Error.
* As gradient boosting is one of the boosting algorithms it is
used to minimize bias error of the model.
9. Example
Age is the Target variable whereas
LikesExercising, GotoGym,
DrivesCar
are independent variables.
Mean Square Estimator
MSE=(∑(Agei –mu)2 )/9 = 577.11
12. Random Forest Regression
* Training a Random Forest Regression
model with our data gave the minimum
Rmse(Root mean square error) of:6.906.
* This gives even better results than the
gradient boosting regression techniques
as it uses the concept of decision trees
and is able to model different kinds of
complex relations that exist in the data
13. Steps for Random Forest Regression
• Pick at random k data points from the training set.
• Build a decision tree associated to these k data points.
• Choose the number N of trees you want to build and repeat steps 1 and
2.
• For a new data point, make each one of your N-tree trees predict the
value of y for the data point in question and assign the new data point
to the average across all of the predicted y values.
there is no interpretability, overfitting may easily occur,
14. Neural Network as Regressive model
Neural Networks Regressive Model using 8
features, with 100 neurons in 1st hidden layer, 50
neurons in 2nd hidden layer and other parameters
as epochs=1000, batchsize=32, Loss metric =
Rmse(Root mean square error), Optimizer =
Adam Optimizer. This model predicted score at
the end of each over with Root Mean Square
Error(Rmse): rmse = 14.21, validation rmse =
14.28
15. Steps for Neural Network as
Regressive model
1. Import the necessary libraries.
2. Import the dataset
3. Build your training and test set for the dataset.
4. Now we have our data we will now make the
model and I will describe to you how it will
predict the price.
5. Now we will fit our dataset and then predict
the value.
16. Conclusion
• In this work we focused on developing a dynamic pricing model for
movie-ticketing business based on demands of the seats for a movie.
• We used a contingent model to predict the optimal demand price based
on the past allocation of seats and the number of seats left for the
show.
• The more data that is collected after instituting a dynamic pricing
scheme, the better results can be achieved. Different kind of models
can be used then for example recurrent neural networks, markov
models etc
17. Real time Example
Demand-based pricing is the highest form of price
optimization. Tickets for movies change over time, based on
the expected demand for each show and the number of tickets
already sold. This strategy is well-known in the airline and
hospitality industry.