Valuable Forecasts Research Articles

From data to harvest: Leveraging ensemble machine learning for enhanced crop yield predictions across Canada amidst climate change

Accurate crop yield predictions are crucial for farmers and policymakers. Despite the widespread use of ensemble machine learning (ML) models in computer science, their application in crop yield prediction remains relatively underexplored. This study, conducted in Canada, aims to assess the potential of five distinct ensemble ML models—Adaptive Boosting (AdaBoost), Gradient Boosting Machine (GBM), XGBoost, LightGBM, and Random Forest (RF)—in predicting crop yields chosen for their ability to manage complex datasets and their strong performance potential. The study integrated various factors, including climate variables, satellite-derived vegetation indices, soil characteristics, and honeybee census data. Data preparation comprised two main steps: first, climate variables were interpolated and averaged for croplands in ArcGIS Pro, along with averaging vegetation indices and soil characteristics. Honeybee census data was also incorporated. Second, the data was organized in Python to create a structured format for models' input. The models' accuracy was assessed using Root Mean Squared Error (RMSE), R-squared, and Mean Absolute Error (MAE). XGBoost emerged as the most accurate model, with the lowest MAE (68.70 for canola and 39.47 for soybeans), lowest RMSE (119.48 for canola and 102.39 for soybeans), and highest R-squared values (0.95 for canola and 0.96 for soybeans) on the test dataset. The study also assessed crop yields under various climate change scenarios, finding minimal variations across the scenarios, but significant negative impacts on canola and soybean yields across Canada. Honeybee colonies were identified as the most influential factor on crop yields, contributing 52.34 % to canola and 57.18 % to soybean yields. This research provides detailed crop yield maps of canola and soybeans at the Census Consolidated Subdivisions (CCS) level across Canada's agricultural landscape, offering valuable forecasts for localized decision-making. Additionally, it offers a proactive strategy for climate change preparedness, assisting farmers and stakeholders optimise resource allocation and manage risks effectively.

Diagnosing Overlapping and Differing Information for SPEAR and CFSv2 Global Precipitation Forecasts

Global climate models (GCMs) provide valuable forecasts of precipitation around the world. This paper has presented an in-depth investigation of the overlapping versus differing information for 2 sets of GCM forecasts based on the classic set operations. Specifically, by using the coefficient of determination to measure the amount of information of precipitation observations contained in GCM forecast, the common part of the 2 sets of forecasts is quantified by the intersection operation and the unique part of 1 set of forecasts is quantified by the difference operation. A case study is devised for the global precipitation forecasts in December-January-February generated by the Seamless System for Prediction and EArth System Research (SPEAR) and the Climate Forecast System version 2 (CFSv2). Their overlapping and differing information are diagnosed. It is found that significant information common to the 2 sets of forecasts exists over 54.61% of global land grid cells, significant information unique to SPEAR forecasts over 23.59% of global land grid cells, and significant information unique to CFSv2 forecasts over 18.15% of global land grid cells. While the information unique to the SPEAR forecasts suggests that the SPEAR forecasts provide new information compared to the CFSv2 forecasts and the information unique to the CFSv2 forecasts suggests that the CFSv2 forecasts also provide new information compared to the SPEAR forecasts, the common information of the 2 sets of forecasts indicates that they present substantial amount of similar information. Overall, the diagnosis of the overlapping and differing information for different sets of GCM forecasts yields insights into GCM predictive performances.

Wishful Thinking or Valuable Forecasts? The Value of Policy Rate Predictions in Sweden

Economists often provide predictions about future interest or policy rates. In contrast to stock market predictions, these forecasts are typically regarded seriously and presented with a degree of accuracy. This paper explores whether placing more confidence in policy rate predictions than stock market predictions is reasonable. It begins by examining historical policy rates predictions in Sweden, comparing them to actual rates to assess their accuracy. The paper then delves into the discussion of how costs of mispredictions, such as sudden and unexpected policy rate increases, can and should be addressed. Despite the Central Bank making predictions within confidence intervals and the banking sector regulating individuals’ borrowing, the burden of unpredicted rate increases falls almost entirely on individuals. The paper addresses whether that is reasonable.

Subseasonal Great Plains Rainfall via Remote Extratropical Teleconnections: Regional Application of Theory‐Guided Causal Networks

AbstractLong‐range U.S. summer rainfall prediction skill is low. Monsoon variability, especially over the West North Pacific Monsoon (WNPM) and/or East Asian Monsoon (EAM) region, can influence U.S. Great Plains hydroclimate variability via a forced Rossby wave response. Here, we explored subseasonal monsoon variability as a source of predictability for Great Plains rainfall. The boreal summer intraseasonal oscillation (BSISO) is related to Great Plains convection and Great Plains low‐level jet (LLJ) anomalies as well as a cross‐Pacific wave train. Using a causal effect network, we found that the time between BSISO‐related geopotential height anomalies and Great Plains rainfall anomalies is about 2 weeks; therefore, BSISO convection may be a valuable forecast of opportunity for subseasonal prediction of Great Plains convection anomalies. More specifically, causal link patterns/maps revealed that the above‐normal weekly EAM rainfall, rather than WNPM rainfall or general geopotential height activity over the East Asia, was causally linked to Great Plains LLJ strengthening and active Great Plains convection the following week.

Design, synthesis and biological evaluation of protease inhibitors containing morpholine cores with remarkable potency against both HIV-1 subtypes B and C

By following up on the design vector of optimizing amine-based HIV-1 protease inhibitors, we have designed and biologically evaluated a novel class of inhibitors with the free nitrogen or sulphone in morpholine cores as P2 ligands in combination with diverse substituted phenylsulfonamide P2′ ligands. As it turns out, a majority of these inhibitors exhibit prominent enzymatic inhibitory activity in low nanomolar ranges with relatively low cytotoxicity. Particularly, inhibitor 1e containing a morpholine carboxamide P2 ligand and a 4-hydroxyphenylsulfonamide P2′ ligand illustrates a robust enzyme inhibitory IC50 value of 90 pM. Furthermore, 1e demonstrates impressive in vivo antiviral activity with EC50 value of 89 nM and a degree of inhibitory potency against the DRV-resistant variant. More importantly, 1e exhibits remarkable activity with EC50 values of 13.59 nM and 8.23 nM against subtype C HIV-1 strains ZM246 and Indie, respectively. Furthermore, the in silico studies provide molecular insights into binding features of inhibitors with HIV-1 protease, and furnish a valuable forecast on further process.

Evaluation of the COVID-19 Era by Using Machine Learning and Interpretation of Confidential Dataset

Various research approaches to COVID-19 are currently being developed by machine learning (ML) techniques and edge computing, either in the sense of identifying virus molecules or in anticipating the risk analysis of the spread of COVID-19. Consequently, these orientations are elaborating datasets that derive either from WHO, through the respective website and research portals, or from data generated in real-time from the healthcare system. The implementation of data analysis, modelling and prediction processing is performed through multiple algorithmic techniques. The lack of these techniques to generate predictions with accuracy motivates us to proceed with this research study, which elaborates an existing machine learning technique and achieves valuable forecasts by modification. More specifically, this study modifies the Levenberg–Marquardt algorithm, which is commonly beneficial for approaching solutions to nonlinear least squares problems, endorses the acquisition of data driven from IoT devices and analyses these data via cloud computing to generate foresight about the progress of the outbreak in real-time environments. Hence, we enhance the optimization of the trend line that interprets these data. Therefore, we introduce this framework in conjunction with a novel encryption process that we are proposing for the datasets and the implementation of mortality predictions.

Forecasting Parameter Strategy Using Data Analytics in Supply Chain Management

The problems of customers on seeking the commodities lead to a strong base for an investigation of Supply Chain Management which necessitates studies on the measurement and evaluation of sustainability. Big Data Analytical approach a hot research area in Computer Science Engineering provides sumptuous scope for handling supply chain management. Constant Support is critical to the creation and maintenance of supply chains competitiveness. Parameters like Demand, Market Values, Customer behavior, weather fluctuations, etc., are to be considered to make out a plan over supply criteria. The previous findings in the area considered now do not include factors like customer preferences, clearing complexity in the process, affordable revenue, time required, natural calamities etc., In this paper, it is proposed to make out an optimal strategy for maintaining the optimal supply chain. A Data set comprising cars influenced by attributes like age, market value, sale price, model, etc., are put into statistical investigation like Regression analysis and compared with Data Analytics algorithms output using feature selection and a contemplation for better prediction of the supply of cars, justifying the effect of the source code and under the realm of Big Data Analytics, a comparison of Big Data analytics is performed to estimate the parameter strategy in the supply chain handling. The Big Data implementation is less time consuming and is devoid of significant fluctuation in deciding the quotation and ordering of the goods. It is a green signal to the reflection of time spirit over the supply chain management. As part of this article, efforts are made to recognize the capabilities of Big-Data applications to make valuable forecasts by establishing a high degree of Reliability and implementing related data. By using the wrapper approach of feature selection, feature extraction is believed to be studied into the prediction model.

Forecasting the Romanian Unemployment Rate in Time of Health Crisis-A Univariate vs. Multivariate Time Series Approach.

Economic crises cause significant shortages in disposable income and a sharp decline in the living conditions, affecting healthcare sector, hitting the profitability and sustainability of companies leading to raises in unemployment. At micro level, these sharp decreases in earnings associated with unemployment and furthermore with the lack of social protection will impact the quality of life and finally the health of individuals. In time of crisis, it becomes vital to support not only the critical sectors of the economy, the assets, technology, and infrastructure, but to protect jobs and workers. This health crisis has hit hard the jobs dynamics through unemployment and underemployment, the quality of work (through wages, or access to social protection), and through the effects on specific groups, with a higher degree of vulnerability to unfavorable labor market outcomes. In this context, providing forecasts as recent as possible for the unemployment rate, a core indicator of the Romanian labor market that could include the effects of the market shocks it becomes fundamental. Thus, the paper aims to offer valuable forecasts for the Romanian unemployment rate using univariate vs. multivariate time series models for the period 2021–2022, highlighting the main patterns of evolution. Based on the univariate time series models, the paper predict the future values of unemployment rate based on its own past using self-forecasting and implementing ARFIMA and SETAR models using monthly data for the period January 2000–April 2021. From the perspective of multivariate time series models, the paper uses VAR/VECM models, analyzing the temporal interdependencies between variables using quarterly data for the period 2000Q1–2020Q4. The empirical results pointed out that both SETAR and VECM provide very similar results in terms of accuracy replicating very well the pre-pandemic period, 2018Q2–2020Q1, reaching the value of 4.1% at the beginning of 2020, with a decreasing trend reaching the value of 3.9%, respectively, 3.6% at the end of 2022.

A framework for assessing the skill and value of operational recruitment forecasts

Abstract Forecasting variation in the recruitment to fish stocks is one of the most challenging and long-running problems in fisheries science and essentially remains unsolved today. Traditionally, recruitment forecasts are developed and evaluated based on explanatory and goodness-of-fit approaches that do not reflect their ability to predict beyond the data on which they were developed. Here, we propose a new generic framework that allows the skill and value of recruitment forecasts to be assessed in a manner that is relevant to their potential use in an operational setting. We assess forecast skill based on predictive power using a retrospective forecasting approach inspired by meteorology, and emphasize the importance of assessing these forecasts relative to a baseline. We quantify the value of these forecasts using an economic cost-loss decision model that is directly relevant to many forecast users. We demonstrate this framework using four stocks of lesser sandeel (Ammodytes marinus) in the North Sea, showing for the first time in an operationally realistic setting that skilful and valuable forecasts are feasible in two of these areas. This result shows the ability to produce valuable short-term recruitment forecasts, and highlights the need to revisit our approach to and understanding of recruitment forecasting.

Modeling Users’ Trust in Drought Forecasts

AbstractForecast valuation studies play a key role in understanding the determinants of the value of weather and climate forecasts. Such understanding provides opportunities to increase the value that users can obtain from forecasts, which can in turn increase the use of forecasts. One of the most important factors that influences how users process forecast information and incorporate forecasts into their decision-making is trust in forecasts. Despite the evidence from empirical and field-based studies, modeling users’ trust in forecasts has not received much attention in the literature and is therefore the focus of our study. We propose a theoretical model of trust in information, built into a forecast valuation framework, to better understand 1) the role of trust in users’ processing of drought forecast information and 2) the dynamic process of users’ trust formation and evolution. Using a numerical experiment, we show that considering the dynamic nature of trust is critical in more realistic assessment of forecast value. We find that users may not perceive a potentially valuable forecast as such until they trust it enough, implying that exposure to even highly accurate forecasts may not immediately translate into forecast use. Ignoring this dynamic aspect could overestimate the economic gains from forecasts. Furthermore, the model offers hypotheses with regard to targeting strategies that can be tested with empirical and field-based studies and used to guide policy interventions.

Anti-Metastatic and Anti-Angiogenic Effects of Curcumin Analog DK1 on Human Osteosarcoma Cells In Vitro.

Osteosarcoma (OS) is a life-threatening malignant bone tumor associated with poor prognosis among children. The survival rate of the patient is still arguably low even with intensive treatment provided, plus with the inherent side effects from the chemotherapy, which gives more unfavorable outcomes. Hence, the search for potent anti-osteosarcoma agent with promising safety profile is still on going. Natural occurring substance like curcumin has gained a lot of attention due to its splendid safety profile as well as it pharmacological advantages such as anti-metastasis and anti-angiogenesis. However, natural curcumin was widely known for its poor cellular uptake, which undermines all potential that it possesses. This prompted the development of synthetically synthesized curcuminoid analog, known as (Z)-3-hydroxy-1-(2-hydroxyphenyl)-3-phenylprop-2- en-1-one (DK1). In this present study, in vitro scratch assay, transwell migration/invasion assay, HUVEC tube formation assay, and ex vivo rat aortic ring assays were performed in order to investigate the anti-metastatic and anti-angiogenic potential of DK1. For further comprehension of DK1 mechanism on human osteosarcoma cell lines, microarray gene expression analysis, quantitative polymerase chain reaction (qPCR), and proteome profiler were adopted, providing valuable forecast from the expression of important genes and proteins related to metastasis and angiogenesis. Based on the data gathered from the bioassays, DK1 was able to inhibit the metastasis and angiogenesis of human osteosarcoma cell lines by significantly reducing the cell motility, number of migrated and invaded cells as well as the tube formation and micro-vessels sprouting. Additionally, DK1 also has significantly regulated several cancer pathways involved in OS proliferation, metastasis, and angiogenesis such as PI3K/Akt and NF-κB in both U-2 OS and MG-63. Regulation of PI3K/Akt caused up-regulation of genes related to metastasis inhibition, namely, PTEN, FOXO, PLK3, and GADD45A. Meanwhile, NF-κB pathway was regulated by mitigating the expression of NF-κB activator such as IKBKB and IKBKE in MG-63, whilst up-regulating the expression of NF-κB inhibitors such as NFKBIA and NFKBIE in U-2 OS. Finally, DK1 also has successfully hindered the metastatic and angiogenic capability of OS cell lines by down-regulating the expression of pro-metastatic genes and proteins like MMP3, COL11A1, FGF1, Endoglin, uPA, and IGFBP2 in U-2 OS. Whilst for MG-63, the significantly down-regulated oncogenes were Serpin E1, AKT2, VEGF, uPA, PD-ECGF, and Endoglin. These results suggest that curcumin analog DK1 may serve as a potential new anti-osteosarcoma agent due to its anti-metastatic and anti-angiogenic attributes.

An explorative study of tourism time series: Evidence from Slovenia and Croatia

Abstract This paper investigates the long-term cointegration between tourism prices and domestic inflation in Croatia and Slovenia. Those two countries share a common economic history and statistical crispness in the 20th century, the time when Econometrics was not a blossoming topic. The two countries split the common economic path in the 1990s and since then, econometricians have been tackling different development issues and researches. The purpose of the paper is to stress the importance of using a well-designed time-series methodology when dealing with multiple variables estimation and evaluation as well in designing adequate and efficient quantitative models, capable to provide valuable forecasts and predict external shocks. It is assumed that, at the basis of an efficient quantitative model, there is a need of unit root and errors normal distribution testing. To test the covariance of cointegration between tourism prices and domestic inflation, the vector autoregressive model (VAR) model is used on 260 valid monthly time-series observations (~ 22 years). The results have shown that prices of short-stay accommodation in Slovenia are cointegrated with domestic inflation, whereas in Croatia there is no stable cointegration vector on prices of accommodation services if / when analysed using the intervention dummy variables and a constant. Although the results indicate that the research hypothesis is generally confirmed, better and more robust results could be obtained including mean-shift dummy variables in a VAR model.

Evaluation of Satisfaction Survey of Decision-Making Meteorological Services in Shaanxi Province, China in 2017

In China, decision-making meteorological services provide meteorological information for the production organization, disaster prevention and mitigation by the CPC committee, government, military leaders and decision-making departments at all levels, as well as scientific decision-making in the areas of rational development and utilization of climate resources and environmental protection. In order to understand the user’s satisfaction with the decision-making meteorological service, the Shaanxi Provincial Meteorological Bureau conducted a statistical survey, and the results showed that: 1) In 2017, the satisfaction level of provincial-level decision-making meteorological services in Shaanxi, China was 92.77%. Among them, the satisfaction index of “Ministry Department Service Personnel Professional Image and Service Awareness” was 94.12%, and the “Weather Forecast Warning Accuracy” satisfaction index was 90.18%. 2) Decision-making users have become an important channel for obtaining meteorological information through meteorological websites, televisions, mobile phone text messages, APP mobile applications, broadcasting, and Meteorological Information Express. Rainstorms, floods, high temperature heat, cold winds, hail, precipitation, and lightning are still the main concerns of decision-makers. 3) The focus on haze and UV intensity is 15% and 8% higher than that of 2016. The next 1 - 3 days weather forecast and 0 - 6 hours short-term forecast are still the most valuable forecast products for decision-making users. Compared with 2016, the next 1 - 3 days weather forecast, future 0 - 6 hours forecast, traffic meteorology, precipitation probability, and air quality forecast increase by 1% to 14% in the year of 2017.

Industry-based equity premium forecasts

PurposeThis paper aims to study whether the industry indexes predict the evolution of the broad stock market in the USA.Design/methodology/approachThe study uses industry indexes to predict the equity premium in the USA. It considers several types of predictive models: constant coefficients and constant volatility, drifting coefficients and constant volatility, constant coefficients and stochastic volatility and drifting coefficients and stochastic volatility. The models are estimated through the particle learning algorithm, which is suitable for dealing with the problem that an investor faces in practice, given that it allows the investor to revise the parameters as new information arrives. The individual forecasts are combined based on their past performance.FindingsThe results reveal that models exhibit significant predictive ability. The models with constant volatility exhibit better performance, at the statistical level, but the models with stochastic volatility generate higher gains for a mean–variance investor.Practical implicationsThis study’s findings are valuable not only for finance researchers but also for private investors and mutual fund managers, who can use these forecasts to improve the performance of their portfolios.Originality/valueTo the best of the knowledge of the author, this is the first paper that uses particle learning and combination of forecasts to predict the equity premium in the USA based on industry indexes. The study shows that the models generate valuable forecasts over the long time span that is considered.

Statistical Deterministic and Ensemble Seasonal Prediction of Tropical Cyclones in the Northwest Australian Region

Abstract Statistical seasonal prediction of tropical cyclones (TCs) has been ongoing for quite some time in many different ocean basins across the world. While a few basins (e.g., North Atlantic and western North Pacific) have been extensively studied and forecasted for many years, Southern Hemispheric TCs have been less frequently studied and generally grouped as a whole or into two primary basins: southern Indian Ocean and Australian. This paper investigates the predictability of TCs in the northwest Australian (NWAUS) basin of the southeast Indian Ocean (105°–135°E) and describes two statistical approaches to the seasonal prediction of TC frequency, TC days, and accumulated cyclone energy (ACE). The first approach is a traditional deterministic seasonal prediction using predictors identified from NCEP–NCAR reanalysis fields using multiple linear regression. The second is a 100-member statistical ensemble approach with the same predictors as the deterministic model but with a resampling of the dataset with replacement and smearing input values to generate slightly different coefficients in the multiple linear regression prediction equations. Both the deterministic and ensemble schemes provide valuable forecasts that are better than climatological forecasts. The ensemble approach outperforms the deterministic model as well as adding quantitative uncertainty that reflects the predictability of a given TC season.

Probabilistic Solar Wind and Geomagnetic Forecasting Using an Analogue Ensemble or \u201cSimilar Day\u201d Approach

Effective space-weather prediction and mitigation requires accurate forecasting of near-Earth solar-wind conditions. Numerical magnetohydrodynamic models of the solar wind, driven by remote solar observations, are gaining skill at forecasting the large-scale solar-wind features that give rise to near-Earth variations over days and weeks. There remains a need for accurate short-term (hours to days) solar-wind forecasts, however. In this study we investigate the analogue ensemble (AnEn), or “similar day”, approach that was developed for atmospheric weather forecasting. The central premise of the AnEn is that past variations that are analogous or similar to current conditions can be used to provide a good estimate of future variations. By considering an ensemble of past analogues, the AnEn forecast is inherently probabilistic and provides a measure of the forecast uncertainty. We show that forecasts of solar-wind speed can be improved by considering both speed and density when determining past analogues, whereas forecasts of the out-of-ecliptic magnetic field [B_{mathrm{N}}] are improved by also considering the in-ecliptic magnetic-field components. In general, the best forecasts are found by considering only the previous 6 – 12 hours of observations. Using these parameters, the AnEn provides a valuable probabilistic forecast for solar-wind speed, density, and in-ecliptic magnetic field over lead times from a few hours to around four days. For B_{mathrm{N}}, which is central to space-weather disturbance, the AnEn only provides a valuable forecast out to around six to seven hours. As the inherent predictability of this parameter is low, this is still likely a marked improvement over other forecast methods. We also investigate the use of the AnEn in forecasting geomagnetic indices Dst and Kp. The AnEn provides a valuable probabilistic forecast of both indices out to around four days. We outline a number of future improvements to AnEn forecasts of near-Earth solar-wind and geomagnetic conditions.

Coupling the short-term global forecast system weather data with a variable source area hydrologic model

Few current modeling tools are designed to predict short-term, high-risk runoff from hydrologically sensitive areas (HSAs) in watersheds. This study couples the Soil and Water Assessment Tool-Variable Source Area model with the Climate Forecast System Reanalysis model and the Global Forecast System-Model Output Statistics model short term weather forecast, to develop a HSA prediction tool designed to assist producers, landowners, and planners in identifying high-risk areas generating storm runoff and pollution. Short-term predictions for stream flow and soil moisture level were estimated in the South Fork of the Shenandoah river watershed. Daily volumetric flow forecasts were found to be satisfactory four days into the future, and distributed model predictions accurately captured sub-field scale HSAs. The model has the potential to provide valuable forecasts that can be used to improve the effectiveness of agricultural management practices and reduce the risk of non-point source pollution.

An integrated ocean circulation, wave, atmosphere, and marine ecosystem prediction system for the South Atlantic Bight and Gulf of Mexico

An integrated nowcast/forecast modelling system covering the South Atlantic Bight and Gulf of Mexico (SABGOM) is in operation, utilizing sophisticated model coupling and parallel computing techniques. This three-dimensional, high-resolution, regional nowcast/forecast system provides a nowcast and an 84 h forecast of marine weather, ocean waves and circulation, and basic marine ecosystem conditions to the public via a Google Map interface. The SABGOM system runs automatically daily and supports a series of user-defined online applications. Extensive model validations were performed online against in situ and satellite-observed ocean conditions. The SABGOM system exhibits a reliable capability of providing valuable forecasts.

Testing the value of probability forecasts for calibrated combining

We combine the probability forecasts of a real GDP decline from the US Survey of Professional Forecasters, after trimming the forecasts that do not have “value”, as measured by the Kuiper Skill Score and in the sense of Merton (1981). For this purpose, we use a simple test to evaluate the probability forecasts. The proposed test does not require the probabilities to be converted to binary forecasts before testing, and it accommodates serial correlation and skewness in the forecasts. We find that the number of forecasters making valuable forecasts decreases sharply as the horizon increases. The beta-transformed linear pool combination scheme, based on the valuable individual forecasts, is shown to outperform the simple average for all horizons on a number of performance measures, including calibration and sharpness. The test helps to identify the good forecasters ex ante, and therefore contributes to the accuracy of the combined forecasts.

Semiconductor industry cycles: Explanatory factors and forecasting

This paper aims to suggest the best forecasting model for the semiconductor market. A wide range of alternative modern econometric modeling approaches have been implemented, and a large variety of criteria and tests have been employed to assess the out-of-sample forecasting accuracy at various horizons. The results suggest that if a VECM can be an interesting source of information, the Bayesian models are superior forecasting tools compared to univariate and unrestricted VAR models. However, for decision makers a spectral method could be a useful tool, which can be easily implemented. In addition, MS-AR models make it possible to obtain valuable forecasts on turning-points in order to adjust the programming of heavy capital and research investments.