Method and system for predicting outbound call

Abstract

The invention relates to the technical field of network communication, in particular to a method for predicting outbound call, which comprises the following steps: S1, data acquisition and processing: acquiring necessary characteristics and target variables including outbound call time, target object attributes and outbound call results from historical outbound call data, and preprocessing the acquired data, such as cleaning, denoising and filling missing values; s2, feature selection and engineering: through a feature selection algorithm, selecting features having important influence on an outbound effect, carrying out feature engineering, and converting and combining original features to generate new features so as to improve the expression ability of a prediction model; and S3, model establishment and training: selecting a proper machine learning model. By analyzing historical data and related factors, a prediction model is constructed to predict the effect and efficiency of an outbound task. According to the method, machine learning and data mining technologies are utilized, and in combination with characteristics of an outbound task and attributes of a target object, prediction is performed and decision support is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of network communications, and in particular to a method and system for predicting outbound calls. Background Art

[0002] In my country, the number of mobile phones has reached 1.3 billion, and the number of fixed-line phones has reached 400 million. Therefore, more and more companies have found that marketing through telephone calls is becoming more and more effective. Telephone outbound marketing has many advantages, such as accurately grasping customer needs, improving marketing efficiency, increasing order conversion rate, unifying sales, and improving brand image.

[0003] Outbound call service, that is, importing a batch of customer numbers for call center agents to make outbound calls, is the main type of call center service, such as the marketing of fast-moving consumer goods such as milk powder, and the sale of real estate, etc. Since the customer numbers imported in batches may be empty or unreachable, if the agents dial them one by one manually, the efficiency is low. Each agent can dial 300 to 400 numbers manually on average every day, and the average connection rate is about 30%, so only 90 to 120 calls can be connected every day. The number of calls that are actually answered (the connection time is longer than 30 seconds) depends on the quality of the number.

[0004] The existing practice is point-to-point outbound calling: one agent corresponds to one dialer. When the dialer detects that the agent is on hang-up, it immediately initiates an outbound call. After the customer answers, the call is transferred to the agent. If the customer does not answer, the dialer hangs up and calls the next phone until the customer answers and the call is transferred to the agent. In this method, the dialer locks an agent, and if the agent is idle, it dials the number to help the agent call the customer. In this process, if the calling customer is not connected, the agent is always in a waiting state, which is inefficient. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for predicting outbound calls to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for predicting outbound calls, comprising the following steps:

[0007] S1. Data collection and processing: Collect necessary features and target variables from historical outbound call data, including outbound call time, target object attributes, and outbound call results, and pre-process the collected data, such as cleaning, denoising, and filling missing values;

[0008] S2. Feature selection and engineering: Through feature selection algorithms, select features that have an important impact on the outbound call effect, perform feature engineering, transform and combine the original features, generate new features, and improve the expressiveness of the prediction model;

[0009] S3. Model building and training: Select appropriate machine learning models, such as decision trees, random forests, and support vector machines, use historical data to train the models, and optimize model parameters so that they can accurately predict the results of outbound call tasks;

[0010] S4. Model evaluation and optimization: Use cross-validation methods to evaluate the performance of the model, including accuracy, recall, and F1 value indicators. Adjust and optimize the model based on the evaluation results to improve its predictive and generalization capabilities.

[0011] S5. Forecasting and decision support: Based on the constructed forecasting model, the effectiveness of new outbound call tasks is predicted; based on business needs and forecasting results, decision support is provided, such as optimizing resource allocation and selecting the best time for outbound calls.

[0012] Preferably, the feature selection algorithm specifically includes defining a search space, evaluating model performance, feature subset search, model training and validation, and model interpretability.

[0013] Preferably, the search space is defined: the search space of the feature subset is defined, recursive feature elimination is used, starting from the full set containing all features, features are gradually eliminated until an optimal feature subset is reached, heuristic algorithms, genetic algorithms, and particle swarm optimization algorithms are used to find the best feature combination by continuously adjusting the feature subset.

[0014] Preferably, the evaluation model performance: for each feature subset, an evaluation indicator is used to measure the performance of the model, and the evaluation indicators include accuracy, precision, recall rate, and F1 score. Appropriate evaluation indicators are selected according to different tasks.

[0015] Preferably, the feature subset search: according to the defined search space and evaluation index, by traversing all possible feature subsets and searching using a heuristic algorithm, the best feature subset under a given evaluation index is found, and the current best feature is selected each time for addition or deletion until the optimal feature subset is found.

[0016] Preferably, the model training and validation: the model is trained using a selected feature subset and validated on a validation set, using cross-validation to accurately evaluate model performance and ensure the generalization ability of the feature selection algorithm.

[0017] Preferably, the feature engineering: after feature selection, feature engineering is performed to further improve model performance, and the feature engineering includes feature scaling, discretization, and polynomial feature construction operations.

[0018] Preferably, the model interpretability: the selected feature subset observes the selected features and observes that some feature factors play an important role in the decision-making of the model, thereby enhancing the interpretability of the model.

[0019] A system for predicting outbound calls, comprising the following modules:

[0020] Data collection and processing, feature selection and engineering, model building and training, model evaluation and optimization, prediction and decision support.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] By analyzing historical data and related factors, a prediction model is constructed to predict the effectiveness and efficiency of outbound call tasks. This method uses machine learning and data mining technology, combined with the characteristics of outbound call tasks and the attributes of target objects, to make predictions and provide decision support.

[0023] The method of predicting outbound calls can help companies optimize the outbound call process and improve the effectiveness and results of outbound call tasks. This method can be applied to many fields, such as marketing, customer service, and research. By accurately predicting the results of outbound call tasks, companies can better formulate strategies, allocate resources, improve business efficiency, and reduce costs. With the continuous development of machine learning and data mining technology, the method of predicting outbound calls has broad development prospects in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0025] Figure 1 A step diagram of a method for predicting outbound calls according to the present invention;

[0026] Figure 2 It is a flow chart of the feature selection algorithm of the present invention;

[0027] Figure 3 The figure is a system block diagram of predicting outbound calls of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] See also Figures 1 to 3 , the present invention provides a technical solution:

[0030] A method for predicting outbound calls comprises the following steps:

[0031] S1. Data collection and processing: Collect necessary features and target variables from historical outbound call data, including outbound call time, target object attributes, and outbound call results, and pre-process the collected data, such as cleaning, denoising, and filling missing values;

[0032] S2. Feature selection and engineering: Through feature selection algorithms, select features that have an important impact on the outbound call effect, perform feature engineering, transform and combine the original features, generate new features, and improve the expressiveness of the prediction model;

[0033] S3. Model building and training: Select appropriate machine learning models, such as decision trees, random forests, and support vector machines, use historical data to train the models, and optimize model parameters so that they can accurately predict the results of outbound call tasks;

[0034] S4. Model evaluation and optimization: Use cross-validation methods to evaluate the performance of the model, including accuracy, recall, and F1 value indicators. Adjust and optimize the model based on the evaluation results to improve its predictive and generalization capabilities.

[0035] S5. Forecasting and decision support: Based on the constructed forecasting model, the effectiveness of new outbound call tasks is predicted; based on business needs and forecasting results, decision support is provided, such as optimizing resource allocation and selecting the best time for outbound calls.

[0036] Preferably, the feature selection algorithm specifically includes defining a search space, evaluating model performance, feature subset search, model training and validation, and model interpretability.

[0037] The defined search space: defines the search space of feature subsets, adopts recursive feature elimination, starts from the full set containing all features, gradually eliminates features until an optimal feature subset is reached, adopts heuristic algorithms, genetic algorithms, and particle swarm optimization algorithms, and finds the best feature combination by continuously adjusting feature subsets.

[0038] Evaluation of model performance: For each feature subset, evaluation indicators are used to measure the performance of the model. The evaluation indicators include accuracy, precision, recall, and F1 score. Appropriate evaluation indicators are selected according to different tasks.

[0039] The feature subset search: according to the defined search space and evaluation index, by traversing all possible feature subsets and searching using a heuristic algorithm, the best feature subset under a given evaluation index is found, and the current best feature is selected each time for addition or deletion until the optimal feature subset is found.

[0040] The model training and validation described: The model is trained using the selected feature subset and validated on the validation set, using cross-validation to accurately evaluate model performance and ensure the generalization ability of the feature selection algorithm.

[0041] The feature engineering: After feature selection, feature engineering is performed to further improve model performance. Feature engineering includes feature scaling, discretization, and polynomial feature construction operations.

[0042] The model interpretability: The selected feature subset enhances the interpretability of the model by observing the selected features and observing that some feature factors play an important role in the model's decision-making.

[0043] A system for predicting outbound calls, comprising the following modules:

[0044] Data collection and processing, feature selection and engineering, model building and training, model evaluation and optimization, prediction and decision support.

[0045] By analyzing historical data and related factors, a prediction model is constructed to predict the effectiveness and efficiency of outbound call tasks. This method uses machine learning and data mining technology, combined with the characteristics of outbound call tasks and the attributes of target objects, to make predictions and provide decision support.

[0046] The method of predicting outbound calls can help companies optimize the outbound call process and improve the effectiveness and results of outbound call tasks. This method can be applied to many fields, such as marketing, customer service, and research. By accurately predicting the results of outbound call tasks, companies can better formulate strategies, allocate resources, improve business efficiency, and reduce costs. With the continuous development of machine learning and data mining technology, the method of predicting outbound calls has broad development prospects in practical applications.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting outbound calls, characterized in that: The steps include: S1. Data collection and processing: Collect necessary features and target variables from historical outbound call data, including outbound call time, target object attributes, and outbound call results, and pre-process the collected data, such as cleaning, denoising, and filling missing values; S2. Feature selection and engineering: Through feature selection algorithms, select features that have an important impact on the outbound call effect, perform feature engineering, transform and combine the original features, generate new features, and improve the expressiveness of the prediction model; S3. Model building and training: Select appropriate machine learning models, such as decision trees, random forests, and support vector machines, use historical data to train the models, and optimize model parameters so that they can accurately predict the results of outbound call tasks; S4. Model evaluation and optimization: Use cross-validation methods to evaluate the performance of the model, including accuracy, recall, and F1 value indicators. Adjust and optimize the model based on the evaluation results to improve its predictive and generalization capabilities. S5. Forecasting and decision support: Based on the constructed forecasting model, the effectiveness of new outbound call tasks is predicted; based on business needs and forecasting results, decision support is provided, such as optimizing resource allocation and selecting the best time for outbound calls.

2. The method for predicting outbound calls according to claim 1, characterized in that: The feature selection algorithm specifically includes search space, evaluation of model performance, feature subset search, model training and validation, and model interpretability.

3. A method for predicting outbound calls according to claim 2, characterized in that: The defined search space: defines the defined search space of the feature subset, adopts recursive feature elimination, starts from the full set containing all features, gradually eliminates features until an optimal feature subset is reached, adopts heuristic algorithm, genetic algorithm, particle swarm optimization algorithm, and finds the best feature combination by continuously adjusting the feature subset.

4. A method for predicting outbound calls according to claim 3, characterized in that: Evaluation of model performance: For each feature subset, evaluation indicators are used to measure the performance of the model. The evaluation indicators include accuracy, precision, recall, and F1 score. Appropriate evaluation indicators are selected according to different tasks.

5. A method for predicting outbound calls according to claim 4, characterized in that: The feature subset search: according to the defined search space and evaluation index, by traversing all possible feature subsets and searching using a heuristic algorithm, the best feature subset under a given evaluation index is found, and the current best feature is selected each time for addition or deletion until the optimal feature subset is found.

6. A method for predicting outbound calls according to claim 5, characterized in that: The model training and validation described: The model is trained using the selected feature subset and validated on the validation set, using cross-validation to accurately evaluate model performance and ensure the generalization ability of the feature selection algorithm.

7. A method for predicting outbound calls according to claim 6, characterized in that: The feature engineering: After feature selection, feature engineering is performed to further improve model performance. Feature engineering includes feature scaling, discretization, and polynomial feature construction operations.

8. A method for predicting outbound calls according to claim 7, characterized in that: The model interpretability: The selected feature subset enhances the interpretability of the model by observing the selected features and observing that some feature factors play an important role in the model's decision-making.

9. A system for predicting outbound calls according to claims 1-9, characterized in that: It consists of the following modules: Data collection and processing, feature selection and engineering, model building and training, model evaluation and optimization, prediction and decision support.

Images