Quantitative market and sales forecast accuracy analysis

The main objective of this master’s thesis was to quantitatively study the reliability of market and sales forecasts of a certain company by measuring bias, precision and accuracy of these forecasts by comparing forecasts against actual values. Secondly, the differences of bias, precision and accuracy between markets were explained by various macroeconomic variables and market characteristics. Accuracy and precision of the forecasts seems to vary significantly depending on the market that is being forecasted, the variable that is being forecasted, the estimation period, the length of the estimated period, the forecast horizon and the granularity of the data. High inflation, low income level and high year-on-year market volatility seems to be related with higher annual market forecast uncertainty and high year-on-year sales volatility with higher sales forecast uncertainty. When quarterly market size is forecasted, correlation between macroeconomic variables and forecast errors reduces. Uncertainty of the sales forecasts cannot be explained with macroeconomic variables. Longer forecasts are more uncertain, shorter estimated period leads to higher uncertainty, and usually more recent market forecasts are less uncertain. Sales forecasts seem to be more uncertain than market forecasts, because they incorporate both market size and market share risks. When lead time is more than one year, forecast risk seems to grow as a function of root forecast horizon. When lead time is less than year, sequential error terms are typically correlated, and therefore forecast errors are trending or mean-reverting. The bias of forecasts seems to change in cycles, and therefore the future forecasts cannot be systematically adjusted with it. The MASE cannot be used to measure whether the forecast can anticipate year-on-year volatility. Instead, we constructed a new relative accuracy measure to cope with this particular situation.

Error Modeling and Accuracy Analysis of a Novel Mobile Hybrid Parallel Robot

Over the last decades, calibration techniques have been widely used to improve the accuracy of robots and machine tools since they only involve software modification instead of changing the design and manufacture of the hardware. Traditionally, there are four steps are required for a calibration, i.e. error modeling, measurement, parameter identification and compensation. The objective of this thesis is to propose a method for the kinematics analysis and error modeling of a newly developed hybrid redundant robot IWR (Intersector Welding Robot), which possesses ten degrees of freedom (DOF) where 6-DOF in parallel and additional 4-DOF in serial. In this article, the problem of kinematics modeling and error modeling of the proposed IWR robot are discussed. Based on the vector arithmetic method, the kinematics model and the sensitivity model of the end-effector subject to the structure parameters is derived and analyzed. The relations between the pose (position and orientation) accuracy and manufacturing tolerances, actuation errors, and connection errors are formulated. Computer simulation is performed to examine the validity and effectiveness of the proposed method.

Cost estimation framework for modular product

Cost estimation is an important, but challenging process when designing a new product or a feature of it, verifying the product prices given by suppliers or planning a cost saving actions of existing products. It is even more challenging when the product is highly modular, not a bulk product. In general, cost estimation techniques can be divided into two main groups - qualitative and quantitative techniques - which can further be classified into more detailed methods. Generally, qualitative techniques are preferable when comparing alternatives and quantitative techniques when cost relationships can be found. The main objective of this thesis was to develop a method on how to estimate costs of internally manufactured and commercial elevator landing doors. Because of the challenging product structure, the proposed cost estimation framework is developed under three different levels based on past cost information available. The framework consists of features from both qualitative and quantitative cost estimation techniques. The starting point for the whole cost estimation process is an unambiguous, hierarchical product structure so that the product can be classified into controllable parts and is then easier to handle. Those controllable parts can then be compared to existing past cost knowledge of similar parts and create as accurate cost estimates as possible by that way.

Sensor Fusion of Proprioception, Force and Vision in Estimation and Robot

Sensor-based robot control allows manipulation in dynamic environments with uncertainties. Vision is a versatile low-cost sensory modality, but low sample rate, high sensor delay and uncertain measurements limit its usability, especially in strongly dynamic environments. Force is a complementary sensory modality allowing accurate measurements of local object shape when a tooltip is in contact with the object. In multimodal sensor fusion, several sensors measuring different modalities are combined to give a more accurate estimate of the environment. As force and vision are fundamentally different sensory modalities not sharing a common representation, combining the information from these sensors is not straightforward. In this thesis, methods for fusing proprioception, force and vision together are proposed. Making assumptions of object shape and modeling the uncertainties of the sensors, the measurements can be fused together in an extended Kalman filter. The fusion of force and visual measurements makes it possible to estimate the pose of a moving target with an end-effector mounted moving camera at high rate and accuracy. The proposed approach takes the latency of the vision system into account explicitly, to provide high sample rate estimates. The estimates also allow a smooth transition from vision-based motion control to force control. The velocity of the end-effector can be controlled by estimating the distance to the target by vision and determining the velocity profile giving rapid approach and minimal force overshoot. Experiments with a 5-degree-of-freedom parallel hydraulic manipulator and a 6-degree-of-freedom serial manipulator show that integration of several sensor modalities can increase the accuracy of the measurements significantly.

De demonstratione exsistentiae et attributorum Dei a posteriori

Invocatio: D.E.(F?)S.

Mobile system for pump working point estimation

In the current economy situation companies try to reduce their expenses. One of the solutions is to improve the energy efficiency of the processes. It is known that the energy consumption of pumping applications range from 20 up to 50% of the energy usage in the certain industrial plants operations. Some studies have shown that 30% to 50% of energy consumed by pump systems could be saved by changing the pump or the flow control method. The aim of this thesis is to create a mobile measurement system that can calculate a working point position of a pump drive. This information can be used to determine the efficiency of the pump drive operation and to develop a solution to bring pump’s efficiency to a maximum possible value. This can allow a great reduction in the pump drive’s life cycle cost. In the first part of the thesis, a brief introduction in the details of pump drive operation is given. Methods that can be used in the project are presented. Later, the review of available platforms for the project implementation is given. In the second part of the thesis, components of the project are presented. Detailed description for each created component is given. Finally, results of laboratory tests are presented. Acquired results are compared and analyzed. In addition, the operation of created system is analyzed and suggestions for the future development are given.

Estimation of existing cable systems in generator-busbar unit and choice the optimal solution for its modernization

The aim of this master’s thesis is to develop an algorithm to calculate the cable network for heat and power station CHGRES. This algorithm includes important aspect which has an influence on the cable network reliability. Moreover, according to developed algorithm, the optimal solution for modernization cable system from economical and technical point of view was obtained. The conditions of existing cable lines show that replacement is necessary. Otherwise, the fault situation would happen. In this case company would loss not only money but also its prestige. As a solution, XLPE single core cables are more profitable than other types of cable considered in this work. Moreover, it is presented the dependence of value of short circuit current on number of 10/110 kV transformers connected in parallel between main grid and considered 10 kV busbar and how it affects on final decision. Furthermore, the losses of company in power (capacity) market due to fault situation are presented. These losses are commensurable with investment to replace existing cable system.

Design and Analysis of Forward Error Control Coding and Signaling for Guaranteeing QoS in Wireless Broadcast Systems

Broadcasting systems are networks where the transmission is received by several terminals. Generally broadcast receivers are passive devices in the network, meaning that they do not interact with the transmitter. Providing a certain Quality of Service (QoS) for the receivers in heterogeneous reception environment with no feedback is not an easy task. Forward error control coding can be used for protection against transmission errors to enhance the QoS for broadcast services. For good performance in terrestrial wireless networks, diversity should be utilized. The diversity is utilized by application of interleaving together with the forward error correction codes. In this dissertation the design and analysis of forward error control and control signalling for providing QoS in wireless broadcasting systems are studied. Control signaling is used in broadcasting networks to give the receiver necessary information on how to connect to the network itself and how to receive the services that are being transmitted. Usually control signalling is considered to be transmitted through a dedicated path in the systems. Therefore, the relationship of the signaling and service data paths should be considered early in the design phase. Modeling and simulations are used in the case studies of this dissertation to study this relationship. This dissertation begins with a survey on the broadcasting environment and mechanisms for providing QoS therein. Then case studies present analysis and design of such mechanisms in real systems. The mechanisms for providing QoS considering signaling and service data paths and their relationship at the DVB-H link layer are analyzed as the first case study. In particular the performance of different service data decoding mechanisms and optimal signaling transmission parameter selection are presented. The second case study investigates the design of signaling and service data paths for the more modern DVB-T2 physical layer. Furthermore, by comparing the performances of the signaling and service data paths by simulations, configuration guidelines for the DVB-T2 physical layer signaling are given. The presented guidelines can prove useful when configuring DVB-T2 transmission networks. Finally, recommendations for the design of data and signalling paths are given based on findings from the case studies. The requirements for the signaling design should be derived from the requirements for the main services. Generally, these requirements for signaling should be more demanding as the signaling is the enabler for service reception.

Loviisan ydinvoimalaitoksen todennäköisyyspohjaisen riskianalyysin tason 2 epävarmuustarkastelut

Diplomityössä tarkastellaan Loviisan ydinvoimalaitoksen todennäköisyyspohjaisen riskianalyysin tason 2 epävarmuuksia. Tason 2 riskitutkimuksissa tutkitaan ydinvoimalaitosonnettomuuksia, joiden seurauksena osa reaktorin radioaktiivisista aineista vapautuu ympäristöön. Näiden tutkimuksien päätulos on suuren päästön vuotuinen taajuus ja se on pääosin todelliseen laitoshistoriaan perustuva tilastollinen odotusarvo. Tämän odotusarvon uskottavuutta voidaan parantaa huomioimalla merkittävimmät laskentaan liittyvät epävarmuudet. Epävarmuuksia laskentaan aiheutuu muiden muassa vakavan reaktorionnettomuuden ilmiöistä, turvallisuusjärjestelmien laitteista, inhimillisistä toiminnoista sekä luotettavuusmallin määrittelemättömistä osista. Diplomityössä kuvataan, kuinka epävarmuustarkastelut integroidaan osaksi Loviisan ydinvoimalaitoksen todennäköisyyspohjaisia riskianalyysejä. Tämä toteutetaan diplomityössä kehitetyillä apuohjelmilla PRALA:lla ja PRATU:lla, joiden avulla voidaan lisätä laitoshistorian perusteella muodostetut epävarmuusparametrit osaksi riskianalyysien luotettavuusdataa. Lisäksi diplomityössä on laskettu laskentaesimerkkinä Loviisan ydinvoimalaitoksen suuren päästön vuotuisen taajuuden vaihtelua kuvaava luottamusväli. Tämä laskentaesimerkki pohjautuu pääosin konservatiivisiin epävarmuusarvioihin, ei todellisiin tilastollisiin epävarmuuksiin. Laskentaesimerkin tulosten perusteella Loviisan suuren päästön taajuudella on laaja vaihteluväli; virhekertoimeksi saatiin 8,4 nykyisillä epävarmuusparametreilla. Suuren päästön taajuuden luottamusväliä voidaan kuitenkin tulevaisuudessa supistaa, kun hyödynnetään todelliseen laitoshistoriaan perustuvia epävarmuusparametreja.

Bayesian methods for estimation, optimization and experimental design

Mathematical models often contain parameters that need to be calibrated from measured data. The emergence of efficient Markov Chain Monte Carlo (MCMC) methods has made the Bayesian approach a standard tool in quantifying the uncertainty in the parameters. With MCMC, the parameter estimation problem can be solved in a fully statistical manner, and the whole distribution of the parameters can be explored, instead of obtaining point estimates and using, e.g., Gaussian approximations. In this thesis, MCMC methods are applied to parameter estimation problems in chemical reaction engineering, population ecology, and climate modeling. Motivated by the climate model experiments, the methods are developed further to make them more suitable for problems where the model is computationally intensive. After the parameters are estimated, one can start to use the model for various tasks. Two such tasks are studied in this thesis: optimal design of experiments, where the task is to design the next measurements so that the parameter uncertainty is minimized, and model-based optimization, where a model-based quantity, such as the product yield in a chemical reaction model, is optimized. In this thesis, novel ways to perform these tasks are developed, based on the output of MCMC parameter estimation. A separate topic is dynamical state estimation, where the task is to estimate the dynamically changing model state, instead of static parameters. For example, in numerical weather prediction, an estimate of the state of the atmosphere must constantly be updated based on the recently obtained measurements. In this thesis, a novel hybrid state estimation method is developed, which combines elements from deterministic and random sampling methods.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.